Steelmanning The NIMBYs

By Scott Alexander

[Epistemic status: very unsure. I sympathize with many YIMBY ideas and might support them on net; this post is me exaggerating the NIMBY parts of my brain to a degree I’m not sure I honestly support. This focuses on San Francisco to make it easier, but other cities exist too. Thanks to Nintil for some of the bright-line argument in part four. Conflict of interest notice: I live in a lower-density part of Oakland]

Everyone I know is a YIMBY – ie “Yes In My Back Yard” – ie somebody who wants cities (usually San Francisco dominates the discussion) to build more and denser housing. This is a reasonable position, and is held by apparently-reasonable people – centrists, rationalists, economists, self-proclaimed neoliberals. Since everyone involved holds reason and civility as an important value, I would expect the discourse around housing to be unusually reasonable and civil.

I have a weird habit of encountering the best parts of some movements and the worst parts of other movements, in a way that doesn’t match other people’s experiences. And certainly I know many YIMBYs who are amazing people who I love. But as for the movement as a whole, I feel like apparently-reasonable people have dropped the ball on this one. Sorry for having to say this, but YIMBYism is one of the most tribal, most emotional, most closed-minded movements I have ever seen this side of a college campus. So much so that even though I agree with much of what it says, I cannot resist writing a 5,000 word steelman of their enemies just to piss them off.

So here are some YIMBY claims and why I cannot be entirely on board with them.

1. San Francisco is uniquely terrible at building new housing

San Francisco currently has just short of 400,000 houses.


Each year, it builds a few thousand new houses, for a long-term growth rate hovering a little above 0.5%.


How does this compare to other cities? I used data from Civic Dashboards to compare the housing stock growth rate of ten major US cities. They only had data from 2008 – 2015, so the analysis only includes those years. They find a higher SF growth rate than listed above, probably because growth has been increasing recently. Here’s what they got:

San Francisco is actually doing pretty okay. [EDIT: Commenter peopleneedaplacetogo points out a chart by metropolitan area rather than city, and using slightly different years, in which SF comes out looking quite a bit worse]

The problem isn’t that SF is building fewer houses than other cities in its league. It’s that demand keeps increasing so much that a normal amount of housing construction doesn’t help.

This might be an unfair objection, because the YIMBY argument might be that San Francisco is uniquely terrible at responding to demand for new housing, and this may be true. But it will important to get a sense for the range of levels of housing construction different cities are capable of, so we can better understand what scenarios are plausible in the next section.

2. Building more housing in San Francisco is an easy way to lower rents

Lowering rents in San Francisco is certainly important: a 1-bedroom apartment costs about $3500. At prices like these, city natives may have to move out because they can no longer afford rent. The lower- and middle- class citizens who work service jobs and maintain infrastructure either disappear or are faced with multiple-hour commutes from the nearest affordable area. Even tech workers with good salaries have to live in overcrowded apartments with multiple roommates to make ends meet. Facets of a good life that depend on having lots of space – like having social gatherings or raising a family – become almost impossible.

The laws of supply and demand suggest that if San Francisco built more housing, the price would go down. This is the foundation of YIMBYism, and it’s basically correct.

But how much would price go down? This requires some economic modeling, which has luckily been done for us.

The San Francisco Examiner follows a paper by Albouy, Ehrlich, and Liu that estimates a 2% increase in housing will cause a 3% decrease in rents. On the other hand, by the time San Francisco has finished building 2% more housing, the population and demand will have increased, meaning that a large portion of the gains will be expended just staying in the same place. They come up with a model that accounts for this, and set themselves a goal of decreasing the average rent for a one-bedroom apartment to “only” $2100 – at least we can’t accuse them of being too ambitious!

They find that this would require a 2.5% housing stock growth rate maintained over twenty years. Going back to the graph from before:

No large US city was able to attain this rate in the eight year period my data comes from, including cities experiencing tech booms during those years. Austin, Texas was able to come close. But at the time, Austin had a population density of 2,500/sqm. San Francisco has a density of 19,000/sqm. Building new houses is easy if all you have to do is clear away tumbleweeds and rattlesnakes – and Austin still only made it up to 2%. We’re expecting San Francisco to clear away existing neighborhoods and angry anti-development activists, and reach 2.5%? And maintain that rate for twenty years?

[EDIT: Commenters point out that the time period my data covers is unusually bad for housing growth. Austin’s records show it was able to grow faster – 4.8% – until 2000, although during that time it was a very small city expanding into mostly empty space. Since 2000, growth has averaged about 2.8%, which is slightly faster than the SF scenario above.]

And even if it works – even if the city can do the impossible – that only lowers rents down to $2100 for a single-bedroom apartment.

Experimental Geography uses broadly the same model, but asks a different question: how much does San Francisco have to increase housing just to tread water and not have rents keep going up? You can read their reasoning at the link, but the answer is “1.5%”:

This seems potentially achievable, but still difficult. A paper by the Federal Reserve finds similarly grim results. I’m not aware of any models that have come to the opposite conclusion.

True, every little bit helps. But affordable housing advocates frequently say that complicated policies like public housing or subsidies are necessary to help poor people who want to stay in high-cost cities. I often hear YIMBYs push new construction as an alternative to these measures, saying that all we need is increased housing supply. But even in the best-case scenario, increased housing supply will take decades to do anything, and lower-income people are at risk of losing their houses now. And even in the best case scenario, increased housing supply will just make apartments slightly more affordable. It’s less likely they can let low-income people live comfortably in the city, or high-income people comfortably raise families there.

3. But at least building more housing will make things a little better, and it certainly can’t make them worse, right?

Devon Zuegel points out that we’re really not sure if that’s true. Why does Manhattan have higher land values than Kansas? Because people want to live where other people (and jobs) are. The denser you make a city, the more other people and jobs will be there, and the higher the land values will get.

Or to put it another way – suppose San Francisco dectupled its housing growth for decades, until it was packed border to border with skyscrapers, and was exactly as dense as Manhattan. In a simple supply-based model, the glut of supply should make rents crash to only a few hundred dollars a month or less. But in actual Manhattan, single-bedroom apartments cost $3800 a month – even more than in San Francisco! If your theory predicts that turning a city into Manhattan will make rents plummet, then consider that turning Manhattan into Manhattan made rents much worse, and so maybe your theory is wrong.

Devon points out that she cannot calculate the coefficients here, so she is not sure whether building more housing will make rents go down (because of supply and demand) or up (because of the Manhattan effect). But we might consider Austin a natural experiment. The model above found that if San Francisco grew housing at 2.5% per year for twenty years, rents would go down by a third. But Austin grew housing by 2.0% per year for about twenty years, and during that time, the average cost of a house doubled. I am not sure San Francisco, which starts from a much higher baseline density, would see the same trend. But at the very least, agglomeration effects suggest all of the terrible and pessimistic models above are still overly optimistic.

Tripling San Francisco’s housing rate until it’s higher than any existing American city, and maintaining it at this rate for an entire generation, might make one-bedroom apartments cost “only” $2100. Or it might do less, or nothing, or make things worse. Right now we don’t know.

4. Holdouts who oppose development are inexcusably selfish, or hate poor people, or are racist

If you want to see real loathing, don’t ask a communist about the rich, or a Trump voter about immigrants. Ask a YIMBY what they think of landowners in a nice quiet part of the Bay who don’t want San Francisco spreading to their area, or who don’t want the BART light rail line connecting their city to San Francisco.

And if you want to see great acting, don’t go to Hollywood or Broadway. Wait for a YIMBY to start monologuing their impression of what these people are like. The exact script differs from person to person, but always includes liberal use of phrases like “the poors”, “brown people”, and “I’ve got mine”.

But I sympathize with these landowners. San Francisco is easy to hate. Even a lot of the people who already live there hate it. They hate the streets piled with discarded needles and human waste. They hate the traffic (fifth worst in the world) and the crime (third most property crime in the US). They hate living five people to a three-bedroom apartment. They hate having aggressive people scream incomprehensible things at them on the sidewalk. They hate the various mutually hostile transit systems that interlock in a system I would call byzantine except that at least you could get around medieval Constantinople without checking whether the Muni and CalTrain were mysteriously failing to connect to each other today. They hate that everyone else in the city hates them, from visible KILL ALL TECHIES graffiti on their commute to work, to a subtle mood of seething resentment from everyone they meet. They hate the omnipresent billboards expecting them to have strong opinions on apps.

I’m not saying everyone in San Francisco hates it. There are people who like all sorts of things. Some people like being tied up, whipped, and electrocuted by strangers. And a disproportionate number of these people live in San Francisco. I am just saying this isn’t a coincidence.

And I sympathize with the people who don’t want BART stations near them. BART stations are also easy to hate. I have a friend who ended up needing stitches after ill-advisedly walking too close to a BART station late at night and getting robbed and beaten up. One of my patients is currently freaking out after their friend ended up in the hospital for the same reason. Some women avoid getting beaten up, but still have stories of getting groped or sexually harassed. BART stations tend to collect a penumbra of litter, drug use, weird people playing incredibly loud music at all hours of the night, weird people shouting at each other at all hours of the night, and the never-dissipating stench of marijuana mixed with urine. This stuff is usually just background noise, but it did make the news last year when forty to sixty teenage thieves took over a BART car in the station and robbed and beat up the passengers, and then again earlier this summer when there were three unrelated murders at BART stations in one week. These don’t seem to have been gang shootouts or anything – they were just people trying to get on their train and getting randomly murdered instead. I am very aware I could get murdered every time I get on a BART. Last time I got off one (three days ago), there was a guy standing in front of the door shouting “FUCK YOU KKK WHITE BITCH” at any woman (of any race) trying to enter or leave the station. Nobody found this surprising or unusual. It’s just what BARTs are like.

But tell a YIMBY that someone, somewhere, is against having a BART station in their neighborhood, and it’s like waving a red flag at a bull.

Maybe clear-cutting everything in the way of San Francisco’s expansion is the utilitarian correct thing to do. Maybe it would increase the US economy so much that we can’t afford not to do it. But Thomas Hobbes wrote that sovereigns may not demand someone go willingly to their death, because resisting death is such a natural human urge that people in the state of nature could not sign it away when forming a primordial state. And some European countries don’t count resisting arrest as a crime, because they consider freedom so fundamental that nobody can be blameworthy for trying to protect it. I believe some people need to have BART stations near their houses, just like some people need get arrested or be executed. But resisting each of these seems so natural and fundamental that I am unwilling to blame anyone for trying. I think neurotypical people usually underestimate how bad cities are for people with noise sensitivities, anxiety, purity intuitions, or just a need for nature and green things in their environment, and that the campaign against people who want to keep their suburbs suburban doesn’t take this into account.

I have heard YIMBYs counter that we don’t have to turn Marin County into San Francisco II, that there’s a balance between trying to preserve what’s good about a place and reflexively opposing all new development. But on slippery slopes where a coalition of people with slightly different preferences are trying to coordinate, drawing a bright line and refusing to cross it is the theoretically correct solution. This is especially true when each new development brings in new voters who may be less attached to the current nature of a place and more willing to vote in future development. Sometimes the only stable solution is just to not get on the slope.

And I have heard YIMBYs counter that if people don’t want to live in an urban environment, they shouldn’t have bought a house in a city. But they kind of didn’t. They bought a house in a medium-density suburb, then some other people came and said “No, this has to be a city”. If they give up, let San Francisco spread to their current home, and move to another medium-density suburb, what’s to prevent other people from trying to urbanize there too? Is our social technology just totally unable to deal with the problem of “how can we let people who want to live in a medium-density suburb live in a medium-density suburb?” Wasn’t the solution supposed to be “these people all gather together, start a community together built to their unique specifications, incorporate, and then pass laws about what their community can or can’t include”? What was wrong with that solution? Some people can’t tolerate the big city – what do you want them to do? Sell their house, leave all their friends and family, and try to start again somewhere else? You think that’s an exaggeration? If where I live became more like San Francisco, I would do that. Lots of people would!

I’m not saying there aren’t compelling reasons to urbanize less-dense areas. But I do feel like there’s a missing mood here that makes me really upset whenever I hear YIMBYs talk about this.

5. Even if building more housing doesn’t lower costs, it will at least allow the people who want to live in San Francisco to do so.

I don’t like the “even if” framing.

If you have been pushing a claim for years, and it turns out you were wrong, then you owe it to the people you have been disagreeing with to say “oops” and take a moment to worry about whether you should lower your smugness level in general. I try to do this when I remember, though I am not always good about it and of course I am limited by my ability to catch and correct my own mistakes.

(I have sometimes been guilty of pushing the “all we need is more housing claim”, so, uh, oops, sorry, upon doing more research I see I was wrong)

The opposite of this is saying “Even if that’s not true, this other thing supports my point”, without explicitly conceding anything at all.

5b. Okay, sorry. Oops, I was wrong about the housing prices. Now that I’ve said that, don’t you also think that building more housing would at least allow the people who want to live in San Francisco to do so?

This is a good and important point, and I think the strongest one in the YIMBY arsenal. I am not really against it, but I can think of two qualms I have with it.

The first is that the argument for ignoring the costs of new construction to existing communities have always relied on the humanitarian necessity of lowering rents for the disadvantaged. If all we’re trying to do is be able to pack a few more people who can pay $3500 a month in, the humanitarian necessity seems less pressing.

But second, bringing more people in helps trap the economy in the same dynamics that caused this problem in the first place.

Some people really enjoy living in dense cities like San Francisco. Other people, for the reasons listed above, really prefer not to. Many of the people who prefer not to are in San Francisco anyway. I signed up to work in the suburbs, but just before I started, my group begged me to work a few days a week in San Francisco because that was where they needed more doctors. I grudgingly agreed. During my time there, I treated depressed San Francisco residents. One refrain I heard again and again was that they hated living in San Francisco, but had come anyway because their company pressured them, or because their companies would pay them extra, or because that was where all the best jobs in their industry were. These people’s long-term plan was to use San Francisco as a springboard to gain enough money or career capital to be able to achieve their dream of leaving San Francisco.

Alon Levy describes the same thing his In The Mines, where he compares the outlook of people moving to San Francisco to that of people working in mines or oil rigs. Nobody likes working in a mine or oil rig. They go there because it pays really well, and if they grin and bear it for long enough, they can pay off their debts or save for the future or do something that allows them to live in a place that isn’t a mine or oil rig:

. People who work on oil rigs, which as a rule are placed in remote locations, get paid premiums. Remote locations with oil have high incomes and high costs in North America, but even the Soviet Union paid people who freely migrated to Siberia or the far north extra. The high wages in this industry are especially remarkable given that the workers are typically not university-educated or (in the US) unionized; they cover for poor living conditions, and a hostile environment especially for families.

I bring up this background because of conditions that I’ve heard second-hand in San Francisco. When I first heard of university-educated adults living several to a bedroom, I assumed that it was a result of extremely high rents and insufficient incomes. But no: I am told a reasonably transit-accessible two-bedroom in San Francisco proper is $5,500 a month at market rate, which is affordable to a mid-level programmer at a large tech firm living alone or to entry-level programmers (or non-tech professionals) living one to a bedroom.

And yet, I’ve heard of Google programmers living two to three to a bedroom in Bernal Heights, not even that close to BART. I’ve also heard a story of people near the Ashby BART stop in Berkeley renting out their front porch; the person sleeping the porch was not a coder, but some of the people living inside the house were.

I have not talked to the people in these situations, only to friends in Boston who live one person (or one couple) to a bedroom, even though they too can afford more. As I understand it, they treat the Bay Area as like working in the mines. They earn a multiple of the income they would in other industries with their education and skills, and have no particular ties to the region. (Some East Coasters have taken to use the expression “drain to the Bay,” complaining that friends in tech often end up leaving Boston for San Francisco.) The plan is to save money and then retire in their 30s, or take a lower-paying job in a lower-cost city and start a family there.

But people have to grudgingly endure poor conditions aboard oil rigs because they’re the only place you can pump oil. Why do they need to grudgingly endure poor conditions in San Francisco?

My understanding is that some industries like technology benefit from centralization. The more programmers are in a city, the easier it is to run a tech company there. The more tech companies are in a city, the easier the job search is for programmers. The more entrepreneurs are in a city, the easier it is to be a venture capitalist there. The more venture capitalists are in a city, the better it is to be an entrepreneur. Add useful infrastructure like Y Combinator and Triplebyte and maybe everyone in tech benefits from being in the same place.

This raises the possibility of a classic inadequate equilibrium, a situation that nobody likes but everyone is stuck in. For example, even if people don’t like Facebook’s privacy policy, interface, or anything else about Facebook, they mostly stick with Facebook because that’s where all their friends are and they’re not coordinated enough to move at the same time. Even if neither passengers nor drivers like Uber, they might use it anyway, because the passengers know that’s where they’ll get a driver soonest, and the drivers know that’s where they’ll get a passenger soonest, and nobody acting alone can break out of the trap.

But if centralization really increases productivity, hasn’t the market decided this is the best solution? I see two ways this might be false. First, it could be that centralization happened in the wrong place – that, if anyone had been able to centrally coordinate, the tech industry should have ended up in Austin or somewhere else that’s well-planned and has lots of geographical room to expand into. Second, it could be that centralization is just a game of keeping up with the Joneses. If there were no San Francisco, then some company would still end up employing the best programmer. But given that there is a San Francisco your company might have to move to San Francisco or have no chance of luring them away from all the companies that have.

Imagine that rising sea levels or something force the evacuation of the Bay Area. Google moves to San Diego, Facebook moves to Santa Barbara, and Twitter moves to San Rafael. Five years later, when Google programmers are sipping daiquiris on the beach in San Diego outside their affordable four-bedroom homes, are they thinking “Man, I wish I was in a crowded unliveable city with multiple inconsistently-connected transit networks right now”? Or are they happy that the option of not living in San Francisco suddenly opened up for them?

The other reason I often hear why people move to San Francisco even though they hate it is because everyone in their subculture is there. Lots of subcultures – queers, hippies, rationalists, etc – seem to be centering in San Francisco. But this might have similar dynamics to the tech situation. Suppose you’re a hippie living in St. Louis, and you’ll be happy as long as there are at least fifty other hippies to form a thriving hippie scene. All the other hippies in St. Louis have some number of other hippies they need to be happy. We can imagine a domino effect where one hippie leaves St. Louis, that causes another hippie to go beneath their threshold and leave St. Louis, that causes more hippies to go beneath their threshold, and so on, until there are no hippies in St. Louis anymore and you have to move to San Francisco or remain tragically un-hip. In this case, the best-case scenario for most St. Louis hippies is that the outflow to San Francisco is limited, so that St. Louis isn’t depleted of its hippie population as quickly. This is also good for hippiedom in general, since there might be proto-hippies in St. Louis who would join the scene if it existed, but who will never convert if all the St. Louis hippies are gone to SF.

(if it sounds like I’ve been thinking about this a lot, that’s because exactly these dynamics have been shaping rationalist communities in cities around the world for the past 5-10 years).

The hyperbolic worst case scenario is that centralization dynamics are too strong, and as more and more people move to San Francisco, life becomes harder and harder for the few remaining stragglers, until finally they give in. San Francisco becomes more and more crowded. Rents increase (through the process mentioned in part 3), number of people per bedroom increases, traffic increases, crime increases. Finally everyone lives in San Francisco, everyone hates it, and nobody can move out – unless they want to give up any chance of working in tech, and spend their entire life talking about cars and football with people named Bud. “San Francisco is unliveable, but at least we’ve made sure lots of people can live there!”

So the counterargument to “Every new housing unit built lets one more person move to San Francisco” is “Every new housing unit prevented saves one person from having to live in San Francisco”.

6. There are no alternatives

I’m not sure this one is wrong.

The argument in Part 5 seems much weaker than the other arguments – so weak that we should probably keep our usual policy of erring on the side of letting people live where they want.

And even if we didn’t want that – even if we thought centralization was a big problem that has to be fought against – it seems weird to leave the fight to crotchety old homeowners worried about noise pollution, and to hope that their self-interest coincidentally creates the world that is best for everybody.

I know there are a lot of urbanists who hate suburbs. I don’t. I grew up in a suburb consistently included in those Most Liveable Towns In The US ranking. It was really nice, and I often remember of it fondly when dealing with the stresses of living in slightly-more-urban Oakland due to me being a dirty rotten defector and participating in the centralization dynamics above. I wish for a world where everyone who wants has a chance to grow up in a nice suburb like that, and I don’t want anyone to have to live in a place like San Francisco unless they’re genuinely into that kind of thing.

I wish for a world with perfect coordination, where half the population of San Francisco decides to move to Helena, Montana at the same time. Half the number of tech companies in San Francisco ought to be enough tech companies for anybody, and the wide sky and endless plains would be a nice change of scenery. I wish for a world where hippies collectively choose Augusta, Maine as the new hippie capital, and so all of the hippies can move there and have great hippie culture and not have to fight with techies for the last $3000 apartment in the Mission.

I’ve heard some people say the federal government should take an active interest in decentralizing tech, since right now one well-placed tsunami could wipe out the United States’ entire technological advantage. I don’t know if this would be a good idea. I’ve heard other people say maybe we can just use virtual reality offices and VR teleconferencing to avoid the need for living anywhere in particular at all. I don’t know if this would be a good idea either.

Since none of those things will ever happen, I don’t know how to get to any of the worlds I want. If there are processes that favor centralization, I don’t know how to fight those processes. I don’t know if there’s some affordable housing policy that would really work. I don’t know if there’s something that balances the interests of every demographic. But I do think that just building more houses won’t, on its own, be a solution to the problem.