Scale matters a lot here, so I'm confused why you want to disqualify the Tokyo example (or Seoul for that matter). Tokyo and Seoul make up a huge fraction of the populations of their respective countries (20-25%). It's true that _locally_ there are agglomeration effects so prices may move up, but if some place is YIMBY enough prices will move down. As far as I'm aware, _every_ locality that builds at the scale that the full-on YIMBYs want them to build as reduces prices.
You may disagree with turning the bay area into Tokyo. But I don't see how if the bay area built enough housing for 66 million people (20% of America) that prices wouldn't go down.
I think there's a significant chance that this is a weird effect of modern zoning (or maybe service economies?) cities used to be where poor people lived! *Downtown* used to be full of poor people! There are major cases of high-density dirt cheap living (like tenements). It seems at the very least plausible that
NYC/SF are expensive because there are MANY good jobs there and people WANT to live there. Not because of the density of housing. You could build 500,000 homes in the middle of your empty field in North Dakota, and it wouldn't do much for the demand there. You aren't going to create Manhattan by magicking 3.5 million housing units of similar quality into the Red Lake Indian reservation in Northern Minnesota.
But overall you are right, that in the real world local density increases can/often lead to local gentrification/rising prices, and the benefits tend to be for the wider area. But the actual impacts depend a lot on the particulars and the services/jobs available.
Take it to an extreme: I have built an apartment in a pocket dimension (TARDIS-style) in New York. It has effectively infinite capacity. What will this do to New York real estate pricing? Do we expect that the apartments next door to mine will become more expensive as a result?
The important thing is to uncouple housing and land. Density increases the value of the land. But it also allows more units to occupy an existing parcel of land, which decreases the per-unit cost of housing.
So here is what you expect to see happen when new units are added at the margin:
1) Cost of all nearby buildings go up (this is actually the land value increasing, but we usually bundle land and building)
2) Value of existing single-family homes increases for the same reason as (1)
3) Cost to move into the neighborhood goes down because there are now more places to live.
It doesn't always happen precisely this way, but this is broadly what you'd expect.
In a perfectly competitive environment, price of housing (not land, the literal house) would be driven down near to what it cost to build. Denser cities usually have more productive people, so labor costs are higher.
Land is also fixed, so that does scale with amenities. Land close to amenities is valued higher, all else equal.
In this theory, how would something like California upzoning everywhere work? Would the local effect drive prices up? (California is one of 50 states) Or would the global effect drive prices down? (All cities doing it together)
Same question if the US upzones. Same question if a city upzones. This local vs global dichotomy seems poorly thought out.
Doesn’t it depend on the link between population and housing? Manhattan isn’t Manhattan because of the number of apartments but because of the number of buildings (and all the services and amenities that follow them).
I doubt the link between the two is as strong as you’d think; there may be a few people who don’t move to a city (think metro area, not urban core) due to house prices, but I suspect they mostly just live in worse (smaller, less convenient, further away, more crowded etc) housing. If you build more housing in Manhattan, you might suck people in from Brooklyn or Jersey City, but I’d bet on it lowering prices in Manhattan’s internal housing market, and Greater New York’s, because you won’t suck more people into the metro area.
In other words, I don’t think housing density causes population density; I think population density snowballs itself, and tends to cause housing density as a byproduct.
Imagine we built 10x houses _everywhere_. What would it do to the land price? I imagine that the total price of all land would remain more of less the same, but the price in the most popular places (Manhattan) would rise because it would be possible for more people to live there. It seems like it implies that the prices in the less popular locations should drop.
Of course I may be wrong about the grand total remaining the same, after all building more housing will create value.
Everyone is bringing up Tokyo, which was my first thought, so I'll instead mention that London is a place that's experienced quite a lot of growth and while it's pretty dense, it's also not very easy to build new stuff there which is why rich people build big underground complexes under their existing houses. For that reason it doesn't seem like a good example of YIMBY not working.
The key empirical facts that go against your story is that physical mobility has declined in the past decade while housing prices have shot up. The tiny share of hyper mobile people you're thinking of aren't nearly enough to move the needle for cities. In general, people move to a city they have no connection to for jobs, not housing.
I think one intuition you're missing here is that allowing for easier construction of housing at the policy level in general shifts housing costs closer to their actual construction cost. And, higher density housing is (to a point) cheaper to construct per unit than sf housing.
I asked an academic about this because I was interested in what the effect taking zoning controls off in areas marked R-60 (single family dwelling) as the local developers and their supporters kept saying it would help develop affordable housing (yet at the same time it seemed like everything new getting built was more expensive than the stuff it was replacing). Locally it does drive up prices. According to my colleague, in the wider DC-metro area it should not because areas in which there were 'old' housing stock have to reduce prices to remain competitive with the 'new' build areas. So Chevy Chase isn't going to get affordable, but mid-South East might see a price drop. At least that's the theory. There's a good example of this in Minneapolis (as they took zoning controls off in 2018). See
I think this is missing an important piece of the puzzle. There's a big potential difference between the number of housing units in a city and the number of people living there. If Oakland had 4.5 million housing units suddenly appear that wouldn't summon an additional 4.5 million people into existence into the city.
The thing that makes New York a more desireable place to live than some island off the coast of Maine isn't the amount of housing there, its the number of people. And not just the people, the pattern of economic activity too. If you created those 4.5 million new houses in Oakland and just teleported people in randomly from across the country they'd all want to move out immediately because they wouldn't have any job to do in this hypothetical Oakland and the things that make a city a pleasant place to live like theaters and restaurants wouldn't exist either without a big shaking out.
If you just created 4.5 million new homes what would happen would be that maybe a few tens of thousands would move in immediately or lets be generous and say 500,000. Then the people who owned the other 4 million would have to slash rents to try to not stay empty and you would see prices plummet enough to start attracting people from all around the Bay Area driving down prices in that metro. But most people have a job or friends of family that prevents them from moving across the country just to take up residence in some apartment that happens to open up.
I think you're making a very common mistake here of confusing supply/demand with *quantity* supplied or quantity demanded. (This is very common! we teach students about this in micro 101 because it's so easy to make!)
What you're seeing is that the quantity supplied is correlated with housing prices (true!). But this is very different from establishing that the supply curve--i.e. the amount of housing that would be produced at any given price, and what moves up/down when we regulate/deregulate supply--is positively correlated with price. Figuring out what supply curves look like is a lot less intuitive and requires some high-grade econometrics, which is why economists had to set up a whole commission just to study this particular problem (the Cowles Commission).
In terms of resources for understanding how these concepts are different, a micro 101 textbook will cover this distinction. For the econometrics side of this, I've heard good things about Scott Cunningham's *Causal Inference Mixtape*, although I haven't personally used it.
There's a number of people who would move to New York if the rent was lower, but it's not an inexhaustible number of people. Keeping housing limited means that the price of housing stays high. The only way to get it down is if supply rises to meet demand. We know that there is an artificial constraint of creating more supply in the housing market in these cities, so it's no suprise that supply is not meeting demand. Every apartment you build is a step closer to satisfying the demand that already exists.
I think that's the issue. The demand is there already, building housing doesn't create demand, it just allows the demand that already exists to be satisfied.
I don't even know what to say further than that. I'm just don't get the induced demand argument. You can't lower the price of something by restricting supply, that just doesn't make sense. London and New York have expensive housing because they restrict supply, not because they encourage it.
I think what one also needs to consider is that $1 is in a sense worth less in NYC than in North Dakota – everything is more expensive in NYC but people also have higher incomes. (I think similarly, 1€ is worth less in Germany than in Latvia. It must be like that in order to balance supply and demand of things from Germany and Latvia, or something like that.)
So, I wonder what that housing price graph would look like if you corrected for that effect.
Price of a house is a function of 2 things: price of land and price of constructing a drywall box with a toilet. Upzoning Oakland definitely wouldn't drive the price of land down, but it would split the land price more ways. So instead of 1 person paying for a 5000 sqft lot, you could have 10 people paying for that same 5000 sqft lot. Even if the price doubles, the price per person goes way down. The cost of construction is higher for building tall skyscrapers vs building single story drywall boxes, but the difference is on the order of 1.5-2x, so as long as the cost of land makes up more than half the cost of your house, then you have margin to decrease that price by building taller. There's probably in theory an optimal density given the supply/demand and construction costs for each parcel and in Oakland, the optimal density is clearly much much larger and best determined by the free market, not city bureaucrats.
It seems to me that the "density increases prices" is not a direct effect, but part of a feedback loop. People are not attracted to more people per se. But to the things those people create (jobs, parties, theaters etc. ). If you just build a lot of houses you won't immediately get these effects, so you can't turn a small village into New York overnight. If you build new houses you get a temporary drop in prices. Untill the new people that live in these houses have allowed for more development to occur (they build a new buisiness or they are customers for the new theater). After that prices increase again. If you then build new houses the prices will lower again. So constant development of houses result in lower prices.
The YIMBY interpretation of the positive association between price and density is that virtually every city in the country adopted a strict zoning regime sometime in the mid 20th century, and the degree to which these laws caused a supply squeeze is mediated by density.
To falsify that we’d need a dense city to drastically liberalize development such that supply spikes upward and see what direction prices move. There aren’t really enough examples of this yet. Outside the U.S. there’s Auckland, which seems to support the YIMBY thesis. But not much that qualifies domestically.
Possible typo: “ But if becoming just as big as Manhattan or London would make Oakland more expensive, shouldn’t we assume that a little step in that direction would make it a little bit less expensive?” Shouldn’t that “less” be another “more”?
You lost me at "So I don’t understand why Matt believes that building a few new apartments in some city - a very small move along that spectrum - would do anything other than make local prices go up." How would they make prices go up? The would just accommodate the latent demand (unless those newcomers have somehow a stronger preference for city housing and willing to pay in excess of existing ones). Of course, one may observe an average increase because the new construction is probably of higher quality/more luxurious, but presumably you want to look at quality-adjusted prices.
I think your mixing up the agglomeration effects of density, which is what induces the demand, and the housing supply. You can't just build a city and expect people to move in, China has tried that. But if you have the agglomeration effects of density and shortage of housing due to artificial constraints, which we have all across the US, then you get dense areas with high housing costs.
Tokyo and NYC both attract tons of new residents But Tokyo's housing rents have been stable, while NYC rents keep rising.
Why? Tokyo has permissive housing construction laws. NYC makes building new housing almost illegal.
Yes, dense cities are attractive, and that makes them get more dense over time. But it only makes them more expensive if you forbid new housing to keep up with the new residents.
Should increasing the supply of housing not induce demand until that supply had been soaked up and the new people have started doing exciting things that induce demand?
So as long as the increased supply keeps ahead of the demand then housing prices shouldn’t go up, but go down by some proportion
There's two things going on here: confusing shifts in demand with movement along the demand curve, and getting causation backwards.
You're assuming density causes prosperity, rather than prosperity causing density. There are ways the former can happen, but the bigger thing is that, for a wide range of historical reasons, you can make a lot of money in NYC and SF, so lots of people want to live there, so they get very dense. This is the prosperity shifting demand right, so at any given price, more people want to live there; this drives prices up, and they go higher the more fixed supply is.
If you built a bunch of housing in Oakland, lots of people would move there because it's cheaper, which is movement along the demand curve; it's still the same number of people who want to live there at any price. Now, it's possible that the increased number of people living there makes the city more prosperous (this is the phenomenon of induced demand), which would shift demand right, but there are way more differences between NYC/SF and Oakland than just the density, so I don't think it would shift demand enough to offset this. In particular, if it's just a small increase in small, it's also a small increase in density, so there's almost no shift in demand (but there is movement along the curve).
Strikes me that the missing factor in this analysis is jobs/industry.
Consider two scenarios: in one (Scenario A), a city permits a bunch of new apartments, but no new offices. In another, (Scenario B) a city permits a bunch of offices but no new apartments.
Obviously the simple existence of an office doesn't guarantee a tenant...but it doesn't hurt (more available offices means it's easier to start a new business). Lots of offices means lots of office workers, means lots of people with professional salaries hunting apartments...I don't think it's crazy to argue that apartment rent is going to go down in Scenario A, and up in Scenario B.
And of course none of these cities appeared from nowhere. You start with a favorable port (usually). Some businesses supporting the operation of the port; then some industry taking advantage of the port; now services for all the people working at that stuff; now corporate headquarters for the growing businesses here. And so on and upward as the network value of the business and industrial cluster goes, achieving true takeoff when you get a New York or London style global financial hub, or a San Fransisco tech hub.
There's clearly a back-and-forth dependence with housing there; the business cluster can't grow if it can't find workers. But I'd argue the housing cost calculation isn't more housing -> bigger city -> more desirable -> higher prices. It's more like industry cluster -> more industry -> more desirable -> higher prices. And critically, that suggests that at any point in that industry network effect takeoff you can add new housing at a rate that will bring down the price of an apartment, because the new apartments aren't the critical factor in driving the business cluster value spiral.
Or to put it another way: NYC isn't the most expensive place because they built the densest housing, NYC is the most expensive place because of the network effect of being the business and financial hub of the richest economy in the history of the world, coupled with regulations on housing artificially restricting access to living near that business and financial hub.
Isn't the Tokyo example just what happens when you exhaust the pool of Japanese people who want to live in big cities? (Or maybe not exhaust -- my guess is that Tokyo is more expensive than random Japanese small towns -- but the gap is not as large as in the US because the fraction of the population wishing to move is smaller). The reason for this is that Japan's population hit a peak at some point in the late 21st century and has since been shrinking; I think if every new Japanese generation was larger than the one before it prices in Tokyo might still be soaring.
You're assuming that cities want to decrease prices. Palo Alto homeowners only care about 1 thing in life: the price of their home. If they could drive up the price of their home by building more, they absolutely would do that. In reality, they know that building more homes would mean more competition for them and also nicer, newer competition. Who would buy a 1200 sqft home built in 1950 when you can get a nice new 2000 sqft townhouse for less?
I’m no economist but it seems there are a few things going on here:
1. Constraining supply when there’s high demand increases prices: this is the main YIMBY argument and I think it makes sense.
2. Currently, dense places are also expensive. Sometimes this is used to argue density doesn’t decrease prices, but surely part of it is about a third variable, namely demand. Again I think this is the standard YIMBY retort to the observed positive correlation between density and prices: nyc isn’t expensive because it’s dense--density and cost are both responses to demand.
3. Demand is driven by a bunch of things, likely including a high density of people (extent of the market, etc). I think this is largely the point you’re making here, right?
I agree that the relationship is probably complex because demand isn’t constant and (possibly) is driven in part by supply. But I think even that could be an oversimplification. People don’t just move to dense cities because they have lots of housing units, they move in the search for opportunity (economic and otherwise). So building more supply per se doesn’t necessarily increase demand: in large part it meets existing latent demand, as per 1 and 2. But to the extent 3 is true, then it’s probably also true that: increasing supply satisfies that demand, increasing number of people, which could in turn increase demand as per increasing extent of the market. Which I think is the feedback loop you’re identifying.
I just don’t think it’s a clean relationship or clear feedback loop though: increasing housing units doesn’t predictably increase demand, presumably--otherwise non-superstar cities like Detroit could just build more housing and would see increased demand and economic growth; but that’s not the relationship we see (afaik). This implies that a large part of demand is independent from (not caused by) housing stock, which is also what a YIMBY would argue, I think. Which means that to the extent supply is limited and demand is high, building more supply is important for satisfying that demand.
But I don’t know. I’m just speculating and this admittedly feels difficult to tease out empirically. It’d be a start to draw some DAGs, though.
Maybe I'm crazy but I thought I remembered an old Krugman paper on the positive network effects of large urban areas, written to explain why the trend of suburbanization was reversing in the 80s or 90s.
Like, there's nice things about SF and the Bay but most people want to be there for the tech scene. If you're a startup, it's worth paying $80k/year for the founder to live in SF because all the VCs and talent is there. Let's say that benefit is worth $100k/year over living in North Dakota. Alright, so if you build 20% more housing and 20% more startups and VCs move in, say decamping from Denver, then regardless of what the rent is, it's now worth $120k/year for the founder to live there because there's more VCs and more talent and more support there.
Like, I'm sure I read this in an old Krugman paper, I cannot find it, but if the primary appeal of a city or location is the network effect for a specific industry and/or general networking effects (like dating options in NY vs North Dakota) then increasing the supply can easily increase the value more than it increases the supply.
> This is a coordination problem: if every city upzones together, they can all get lower house prices, but each city can minimize its own prices by refusing to cooperate and hoping everyone else does the hard work.
There's a much worse problem involved: everyone who already owns property in a city doesn't want the property prices there to go down. "The city" might want that, whatever that might mean, but individual home owners don't. Nobody who bought a house for $1,000,000 wants it to cost $500,000 after the housing supply in the area doubles. All those NIMBY talking points about the crime and whatnot mask the real problem: they don't want to lose money because their house price decreases.
This is not a coordination problem and you got the incentives exactly backwards.
I don't see an ethical way of solving this problem, in a way that that isn't like, "I want to move into your city but the prices are too high so the people in your city should have half of their wealth wiped out so that I, not a resident currently, could move in".
That is, sure, building a megacity on the plains of South Dakota would work and then the property owners in other megacities complaining would look just as ethically unfounded as people complaining about NIMBYs. I think that it's a good idea.
You're missing the basic tradeoff of a city: the closer to the central business district, the higher the prices. The bigger the city in total population, the higher the price in the central business district, because there are more people fighting for that desirable place with the best transportation network access to everywhere else. This central area is expensive because of demand, but also because of physical constraints, skyscrapers are inherently more expensive than wood-frame buildings. The higher the skyscraper, the more costly it is. New housing will never make Manhattan affordable.
Where there is room for affordability is on the periphery and on transportation lines going into the central business district. This is the main lesson of Tokyo. One might look at Tokyo's central business district's prices and tiny expensive apartments and conclude market urbanism is a failure to achieve anything. But along the train lines that go into Tokyo, things get rapidly more affordable. YIMBYs should still support bigger skyscrapers in central business districts, but they should only look for affordability further out in 6 story buildings within walking distance of high-quality public transit lines.
So if NYC upzoned single family homes in the periphery burrows and improved transit, housing in Manhattan would probably get more expensive, even as there was increased affordability for the area from which one could commute into Manhattan. It's possible in the short run we might have uneven growth in demand that could continue to outstrip supply. Trains can move lots of people, cheap 6 story buildings can accommodate lots of people, and even if the ocean blocks expansion in one direction, that still leaves a half circle that grows the area people can live in quadratically. There's lots of capacity to meet demand to live within commuting distance of a desirable city center if there's political will to build transit and allow housing.
Edit: After reading other responses, I think I might agree. It's unreasonable for YIMBYs to expect small increases in housing supply to necessarily decrease short-run prices. Achieving affordability would be a long-run project. Short-run expectations should be heavily managed.
Density is not a destination or a characteristic of an area (and it's unfortunate we talk about it that way in Yimbyland).
- The more pedantically correct Yimby claim should be "increasing supply elasticity in the housing market, via regulatory reform, will reduce housing prices relative to local incomes and blunt price increases during economic expansions as well as set up loca housing markets for bigger price reductions during recessions (because we let developers build themselves off a cliff at the top of the market)."
Density doesn't drive prices.
- Prices drive density (within the constraints of the local land use regime), and prices are driven by demand. Yes, if we allowed denser construction, which definitionally would mean a reduction in per capita land consumption, ///relative/// prices would decline and more people could move in to a high demand area...but now we're talking about more people moving in to a high demand area ///because of the decline in relative prices///.
All cities are not perfect substitutes for each other for prospective residents.
- There surely some people doing the digital nomad thing or choosing which McKinsey office to work from in their late 20s, but I see no evidence that this population is large enough to have any appreciable impact on housing prices. Folks' personal and professional networks aren't perfectly distributed across all major cities.
What Scott is missing is that "density" isn't one thing. To a first approximation, it's comprised of JOBS and HOUSING. These are complements. If the amount of JOBS density increases, the price of HOUSING increases. If the amount of HOUSING density increases, holding JOBS constant, then the price of HOUSING falls.
This isn't at all counterintuitive and comes out of the canonical Alonso, Muth, Mills model (although without my cartoonish names).
Economics PhD student here, and first comment here.
I think Scott's basic framework is mostly right, but could use some tidying. His starting graphs basically depict a supply curve -- a positive relationship between prices and quantity, because if prices are higher then developers build more houses. But he's confusing a shift of demand along a supply curve with a shift of the supply curve itself.
Here is what I think Scott is trying to say. There is a simple supply and demand model that YIMBYs appeal to, with a simple result that pushing out the supply curve increases quantity and lowers prices, hooray. Scott is saying that there is a knock-on demand effect, what he calls "induced demand" and what activists call "gentrification" and what economists call "endogenous amenities".
This is a thing that definitely exists, but the relative sizes of the effects will determine what the end result is, and we also have to bear in mind that development occurs in response to changes in price. And the studies that Yglesias points to are essentially concerned with precisely disentangling and measuring these effects. The results generally say that the supply and demand curves are such that the price ends up lower overall, due to a combination of the slopes of the curves and the sizes of the shifts. We frequently do observe increasing prices and development together, but this is because the increase in prices causes development, which then (as the studies show) causes prices to be lower than if development had not occurred.
So that is why density does not cause price increases -- increases in price cause development, which causes a counterfactual reduction in price, which outweighs the effect of the endogenous amenity of density.
I'm not sure that the demand-increasing effect of density is big enough to make a substantial difference in practice. E.g. if lots of housing, enough to constitute a large fraction of the total, were built quickly in an already large city, then the opportunities arising from density would go up, & probably so would demand. But on the margin, the effect of each unit of housing on the opportunities arising from density is minuscule, so that that marginal upward effect on prices would probably be outweighed by the downward effect of normal supply & demand. (& of course, if a similarly enormous amount of housing were built quickly in a small city, it probably wouldn't cause demand to increase in the same way because there would be no guarantee that enough people would move there to generate the opportunities produced by greater density.)
My guess is that price as a function of housing supply would follow a relationship along the lines of Price=M*Quantity+B, where M is negative but B increases enough to offset that if the population density goes up by a lot (this depends on other things besides quantity supplied, like the local economy, so that New York, a typical small city, & the empty field in North Dakota can't be modeled as two points on the same line).
Good piece. As a longtime gentle YIMBY skeptic, I've always been a little bemused that the same crowd insisting that the rental market is 'just supply and demand' would be horrified if you said 'immigration obviously reduces wages, don't you know about supply and demand'. They would rush to explain to you that no, wages are much more complex than an Economics 101 understanding of supply & demand, that the new immigrants create demand for new goods and services and so actually create more jobs by their presence, and so on. I would personally agree with this pro-immigration argument, I just think it's funny that to the YIMBY crowd simplistic 'supply & demand' explains real estate prices but doesn't explain wages.
Cities are huge agglomerations of network effects- they're valuable precisely because there are more potential employers, romantic partners, friends, business connections, activities and more- it's quite possible that a higher population makes the city more valuable and so raises rents! (And wages). I think Devon Zuegel had a post about this from a few years ago. Anyways, 101ism is a bad way to understand complex systems
I think the basic problem is that you're ignoring other covariates in your regression, most importantly the demand to live in a city. As a result, the regression coefficient is incorrect. This is a well-known problem in statistics, called the omitted variable bias (https://en.wikipedia.org/wiki/Omitted-variable_bias). I believe if you account for the other variables then the regression coefficient will be negative as expected from economic reasoning.
This is why marginal analysis is important in economics. While comparing two situations many variables will be different, so it isn't clear exactly how the outcome is related to one variable in particular. In marginal analysis the two cases are pretty much the same and the only difference is the marginal change in one variable. So any change in the outcome can be attributed to that variable.
I think the answer to this hugely depends on your internal mental flowchart. This post seems to suggest a model of:
housing -> cool place -> high prices
Which I think buries the step that is:
housing -> support for cool amenities -> cool plate -> high prices
This in turn allows for some finer grained analyses: If Norman, OK (the lowest population density city in the US) built even a moderately large amount of housing, it probably would crash housing market there - no matter how much housing there is, it's not going to experience a housing boom. On the other hand, Austin TX strikes me as too far down the path of becoming a "cool city" to avoid price growth by building housing. The marginal person probably avoids moving to Austin at this point because it's too expensive for the existing amenities, rather than because there's not enough to do there. And finally a city like NYC or SF is probably already "too cool" to realistically discourage interest by preventing construction.
Oakland then becomes a bit of a weird case - it's close enough to a major high cost area that if it got cooler it really might start growing new amenities (restaurants, better (perception of?) public safety, better public transit, more people you know, etc.) and the increased amenities might in turn drive increased demand. But it's not really cool enough that its growth is locked in, at least as far as I can tell.
I think the "ground truth" here is probably somewhere between your and Matt's position - construction really seems to have a local, short term, effect in raising prices; construction is a key element in becoming a successful big city, which is correlated with higher prices; and the exact trajectory of a city between "low cost, mid sized city" and "high cost megapolis" is probably determined by some amount of coordination problems that are fairly opque.
I don't think it's obvious that there is any agglomeration effect from the data presented.
Yes, there are some Americans who want to live in a big city. This is part of what makes New York more attractive than Lubbock, TX for a random 30 year old. But your charts make a comparison of density within a class of "big US cities". Do you really think people consider Houston and SF, and think "yeah I'm going to live in SF because it's really dense. I know it's half the population, I just want the density." It doesn't pass the common-sense test.
The reason people move to SF and NY is because of jobs and culture, not specifically how scarce housing is. In fact, that is an obvious reason people don't move to those cities. The culture/jobs demand drives up house prices, and consequently makes cities more dense. But if you just build more houses, quantity supplied is rising due to strong demand. There's nothing magic about density itself. If you build more houses, you have a shift right in supply. This lowers prices.
It's a fair argument to say, "well why hasn't this worked anywhere in America?" I guess the argument would be, it hasn't really been done on a sufficient scale, so any supply-side effects are overwhelmed by demand-side effects. But I'm not confident in this. Just not convinced that building houses has any positive causal effect on local house prices.
> They moving there because they want to be in a big city - with friends, jobs, museums, and nightlife.
Ok, but these things are not synonymous with density. In fact, having more office space, museum space, restaurant and nightclub space, etc. means having *lower* population density, ceteris paribus, because you can't live in any of those other types of building. People would still want these things if there were no big dense cities. Either they wouldn't exist, or people would still try to live in the places where those things are available, which be more expensive than NYC is today. Basically, I think that the effect of density -> amenities is much smaller than the other way around. Cities are built for economic reasons--you say " Manhattan has a few extra natural amenities, like a river and a good harbor. But nobody moves to Manhattan for the harbor. They moving there because they want to be in a big city - with friends, jobs, museums, and nightlife." but this isn't really true. The harbor is why the city exists at this location in the first place. It's no longer the direct reason why NYC is an economic hub, but its current major industries (finance, research, media, tech, etc.) are there in large part because it was previously a center of immigration, trade, and manufacturing.
Building more housing would not turn Oakland into SF or NYC.
Consider also that the population of Manhattan peaked around *World War 1* (https://en.wikipedia.org/wiki/Demographics_of_Manhattan) at around *40%* more than today. Do you think it was cheaper to live in back then, or more expensive? Given the average salary of a factory or port worker, I'm going to guess cheaper.
>. But doesn’t induced demand violate the economic law of supply and demand?
> if every city upzones together, they can all get lower house prices, but each city can minimize its own prices by refusing to cooperate and hoping everyone else does the hard work.
I don't think this is what the data really show. At this point, building housing in the expensive areas would just serve the already existing amenities that are already drawing people there. And it seems like it's a fairly common development pattern in the US for suburbs to be more expensive than the city they surround. But also, many towns don't *want* their own housing prices to be cheap. "Diminished home values" is one of the most common NIMBY talking points against, well, everything. If you own a home, then expensive housing is good for you, and bad for everyone else.
Ok, really silly. We are comparing the overall "life opportunities" of North Dakota and NYC and then wondering why housing is cheap & low density in N Dakota? Really? Maybe application of common sense should come first before writing.
Sorted by population density, NY #1, Jersey City New Jersey is #2, and Paterson New Jersey is #3.
Newark is in the top 10, and Philadelphia and Yonkers are in the top 15.
So your thesis appears solidly disproven.
People either want to live in city X, or they want to live in a small set of cities Y. They do not want to live in "any city Z with sufficient population density"
Rather than disagreeing with the conclusion, I want to object to the whole framing of focusing on the effects on housing prices.
It’s not immediately clear what police goals should be with respect to home prices. Existing homeowners who think of their homes as investments would like prices to go up; renters and new homebuyers want prices to go down. Numbers on both sides of this issue are close enough that neither position is a political slam-dunk, and nobody on either side is going to change their mind easily (unless they buy a house I guess), because they’re all voting in their own economic interests.
So the end result of focusing on home values is always going to be political paralysis, and framing the issue in those terms already gives the game away to NIMBY forces that profit from political paralysis.
I don't have strong opinions on the details, but I would be cautious about false assumptions of linearity. Part of what makes places like Manhattan and London expensive is the network effect, which makes them desirable. Density is one variable that contributes to their network effect, but:
1) it is not the only such variable, and
2) even only considering density, network desirability is not a linear function of density.
It doesn't seem crazy to me that, due to other variables, add a little bit of housing, in a way that doesn't much change network effects, could move prices in the opposite direction from adding a great deal of housing, that did.
"I don't think there are enough immigrants to really affect things."
Given that 31-39% of the Bay is foreign-born _right now_ and the global population of mobile individuals ready to leave their homes to work for a wage lower than prevailing Bay Area wages is at least a billion, maybe more... how can it not affect things? Net immigration is in the millions per year to the US... all of them by definition need somewhere to live... how the Bay vs. the US. vs. the World a self-contained proposition?
I think we might agree to a rough model of "Demand = Price + Supply", and lets first consider what happens when demand is fixed: high demand expresses in high price and supply. Randomly generating some cities with high and low demand, you'll see high price and supply co-occur, as in your correlations. But once these cities are made and their demand is fixed, a hypothetical urban planner sees price and supply trade off each other, so increasing supply is how you decrease prices.
Now obviously demand is not fixed, and supply up -> price down -> demand up could happen (and I would wager does)! But what is the coefficient? If X>1 and each new home makes more than one person want to additionally live there, then building new homes raises prices, while for X<1 econ 101 holds and increasing supply decreases demand. This is precisely where you need to look at the empirical literature to measure X, and it says upzoning lowers costs.
I'm looking out my window at the crowds gathered to catch a glimpse of the celebrities arriving at the Met Gala in a few hours.
No matter the number of new housing units built, would Oakland ever have something like the Met Gala or in the case of LA, the Oscars? Maybe in a hundred years. It takes a great deal of time and luck to become a New York or an LA.
As for Miami, it's a contender, but way too early to tell.
1. This is why I'm so excited about the recent YIMBY bills passed at the state level in California. I think a change at the state level is the only way to move forward instead of hoping that each NIMBY city changes their minds individually.
2. This argument contradicts the stereotypical NIMBY reasoning of "if you build that apartment building 2 blocks away, it'll add more supply and the value of my home will go down". They should support housing, because the value of their property might actually increase!
3. If there's really people out there who are willing to choose "any big city", shouldn't these people move to the other cities as prices keep going up and up in NYC and SF? Maybe people and companies will start moving to Austin, Miami, Atlanta, Cleveland, Detroit, Sacramento, Minneapolis, etc. Honestly SF and NYC are not worth their high price *right now* in my opinion.
I think you are in general right that agglomeration effects are real, which is why bigger cities have higher value to residents. I agree that people move locations. But I think you can go a step further. If one city is growing faster and densifying, surely those people are not demanding homes in other cities and those cities build slower. This is part of the spatial equilibrium story that further makes claims about “build density and get cheap homes” less plausible.
It is not reverse causation but rather the existence of a confounder - the attractiveness of the location. Imagine cloning that island off the coast of Maine and gluing it to Manhattan. Many people will want to move to the clone, but few will want to live on the Maine island. 20 years from now, after developers have built housing on the NY island, it will be both denser and more expensive than the Maine island.
Also, don't forget that demand is the willingness AND ability to buy. Those dense cities also have lots of well paid people. That increases demand and drives up prices. You need to control for income before concluding that denser places are more expensive.
Edit: Point 3: I just noticed that the Y axis is pct of income spent on housing. That will also be affected by income. Someone who takes home 10K after taxes will struggle if he pays a third of his income for housing. Someone who takes home 50K can afford to spend half of it on housing and still have plenty left over.
Popularity breeds popularity; that's a "network effect," and it holds for cities just like for music concerts or second languages.
But does popularity breed prices? That depends. Scalped tickets to hot concerts are super-expensive; learning English isn't.
Both "tickets to the popular concert" and "English proficiency" have network effects, but only one has limited quantities available. The price effect is obvious.
American cities, and most cities worldwide, limit the quantity of housing. So in most cities, popularity acts like concert tickets: density goes with high prices.
A very few cities like Tokyo don't limit the quantity of housing. So there, popularity acts like English proficiency: density goes with flat prices.
I think there are both amenity and disamenity effects from density, but the real thing that pushes up housing costs is having lots of jobs nearby. To some extent, jobs will tend to cluster near housing, but both job decentralisation in the post-war era and re-centralisation since the 90s show it mostly happens independently of housing growth in this our very regulated market.
I think this is missing something by categorizing a city as entirely "lots of people (and their housing)." You can do better by modeling cities as containing 2 categories of stuff: housing and attractions. People want to move to a city because of attractions. Building more housing raises the number of people able to live in a city without increasing the number of people who *want* to live there--possibly, in fact, decreasing it (because people want to live in areas with high attactions per (area, resident, etc)) In this model, there are 4 different ways you can move--increase housing, decrease housing, increase attractions, decrease attractions--each with obvious effects on pricing.
So, really, if we want cheaper housing in (say) NYC, we should bulldoze central park, close all the museums and restaurants, and ban broadway plays.
I am so excited that you've finally thought yourself almost all the way around to becoming a NIMBY like me.
The final bit you're missing, I think, is that the best solution is to build a bunch more housing everywhere _except_ the biggest metros. The San Francisco Bay Area has eight million people, which is too many, and its geography is amazingly poorly suited for a giant metro area (what with the huge bay in the middle, and the steep hills, and the earthquakes).
We need to find ways to stimulate growth in cities in the 100K-1M range, and maybe the 1M-3M range, without cramming more people into the small number of overcrowded megalopolises that already exist.
It matters what is being built. People don’t move to NYC because there are so many houses there. It is the industry and culture. If you are building more things like stock exchanges and museums, housing prices will go up at a constant housing density. That doesn’t mean building more houses will make housing prices go up.
I think the unstated assumption is that size in and of itself is what attracts more people to a city and hence increases demand and prices, rather than something unique about the cities, like architecture or neighborhoods. It’s not intuitive to me that New York sized Houston becomes popular just because it’s big. (This hints that other features that made cities desirable started the flywheel.)
The "induced demand" argument essentially says that there are a significant amount of people who don't want to live in area X now, but would if it were a bigger city. I can kind of see the argument for this in some cities, but who could possibly be thinking this about NYC and the Bay Area? Those are already *the* places to be - building more housing can't make them any more so.
Also, I'm going to nitpick the density-vs-pricing graphs you picked. The first one is weighted by a bunch of unlabeled data points on the lower end, and the second one shows a frankly unimpressive R-squared value.
Off topic I know but I can not find the SSC post talking about how scarcity leads to politics and scott makes up a fictional example of if there was 10% of the available water all the interest groups would fight for it but with enough water it is a nonissue. It could be a ACT post but I think its older.
I think the hypothesis isn't clearly defined, and the question is phrased in a way that makes a correct answer impossible. If building more houses causes local prices to go down for 4 years, and to go up from then on, did it make house prices go down, or up? If building more houses causes others to build more amenities, and attracts more businesses, creating more jobs... how far down that chain can we go, and still attribute raise in local house prices to building more houses?
Each of those additional steps--building museums, creating new businesses, etc.--costs additional money. It seems to me that the credit for the rise in housing prices should be distributed among the different factors proportionally to their capital (or other) investment.
You're approaching the question like a foundationalist, who believes that assigning credit for a result to a single cause is unproblematic. Correct epistemology must allow the dynamics of recurrent causal networks to be computed iteratively, and an answer to be given in a form that could reproduce dynamics like "price goes down for 4 years and then goes up".
It might be interesting to look at the Spanish market for a possible counterpoint.
Spain has a very uneven population, with most concentrated in the coast and the center. The sparse parts of the country are very sparse, and the dense ones are very dense.
In the rally to 2008 Spain was building A LOT of new units [1]. Prices were rising at a very rapid rate. At the same time population was growing fast, mainly due to immigration [2].
When the bubble burst in 2008, home prices started going down[3]. Not surprising considering the inflated prices and the massive uptick in unemployment [4]. However what's surprising to me is that this price slump lasted a really long time, and only recently we are approaching prices close to the top of the bubble. Is is true that population growth slowed down a lot, and for a few years between 2013 and 2016 total population actually went down. However median salaries [5] sort of grew steadily over the period, again, not accounting for unemployment. Still, the bottom of prices seems to have happened around 2016, when unemployment had been decreasing from it's peak for a few years.
Of course there are multiple interrelated causes happening at once in this case, but my take is that high density helped to tame housing price for a relatively long time, even as economy recovered from the recession.
I think a lot of the studies showing building reduces costs locally are accurate in the short term, but the reason you're right is that it takes time for the new supply to attract out-of-area residents, increasing the population, and increasing desirability. A lot of this can be a multi-year cycle, i.e. New York builds some housing, on the margin your friends want to move there, they do, 3 years later a critical mass of them is there, and you move there as well. And then at this point, it actually did contribute to a raising price.
I think what you and Matt are both saying could be consistent. For example, it might well be the case that on the margin adding more housing to New York doesn't increase its "New Yorkiness", i.e. its having of Broadway, Wall Street, and so on, because those are features of being the #1 city, not features of being a city of N million people. In this model there could be discontinuities when you get enough extra people to cause some sort of runaway migration to your city, but it wouldn't increase demand more than supply on the margin.
Of course this could totally be wrong, and it could be that with 5% more people New York would _also_ become the center of software in the US or something.
A topic close to my heart - and I suppose many other's too. Thanks for this. I am EXTREMELY skeptical of any claim that building more homes lowers prices or eases affordability for families. I live in an area that built tens of thousands of new units and rents in these buildings basically monotonically increase.
There are a few main factors as I see it (well, maybe it's really just one):
1. Living in a globalized world. More houses for sale is just more inventory from which wealthy domestic and international buyers can select from. If there is an ephemeral drop in price, it will entice more of these buyers until the situation returns to exactly as it was before.
2. Everyone in the housing market (buyers and sellers) have wisened up and are playing the same game. With the explosion of equities and real-estate following the zero-interest rate environment for so long, people's net worth blossomed. Though equities are down a little, most have now wisened up and have moved that wealth into money market and related bond funds that are earning them a nice 5-6% interest rate. This means lots of people have lots of money - in my area development just keeps going further out and, even in April 2023, open houses are packed and houses sell after 2 days for 100s of K over asking - often in cash. Just as before, if creation of new housing temporarily lowers price due to a higher supply, it invites an even larger horde of potential buyers who are in the (n-1)th tier of wealth - and then it returns to the equilibrium of before, where everything beyond the affordability of almost everyone, yet any house on the market (unless utterly dilapidated). Building more houses just invites more buyers from more places -- and rinse and repeat.
3. Building new housing (particularly in wealthy, economically dynamic areas) increases the population and with that comes more business, jobs, investment and all the things that make a place desirable and expensive. The ultimate attractor of the system is to be unaffordable. If it was affordable, everyone would buy it all up until it was unaffordable all over again. I think the logic is as simple as that. Maybe a wealth tax? And/or a severe recession where people lose jobs and wealth and HAVE to sell are the only things that would change this.
I think the YIMBY argument is that it's interesting that altering the amount of housing in an area, resulting in higher density in that area, doesn't result in higher prices. You seem to be saying "ah yes, but under our current planning laws, more people moving to a city does increase prices in that city".
This seems like the same argument seen from different angles. You're making the point that in the past, prices rose with density on a city-wide basis, even if that doesn't hold for small areas within a city. Which sounds like something YIMBYs would agree with.
I think the response might be "yes, the point of the small scale studies is that they suggest it's not increased density itself that increases prices. It's something else, something that only shows up at a larger scale, over a longer time period, and we think that that something else is the overarching planning regime".
And that's where Tokyo comes in - you can say, look, there's another wealthy, developed country with far less restrictive planning laws, and in that country increasing density doesn't seem to lead to increasing prices.
If you've got a vetocracy, increasing the number of people with vetos will probably decrease the number of laws passed. But that doesn't mean that it is inevitable that larger populations are more inclined to paralysis. It means you have to change your system of government.
I'm skeptical about this model at the city level but I think there's some truth to it at the micro-neighborhood level. Building ten new homes on my block would not meaningfully increase affordability of my immediate area, since this block is more or less fungible with dozens or hundreds of other blocks surrounding it. If anything, it makes the micro-neighborhood slightly "fresher" and should increase property values in the immediate area (for example, because those existing properties are now adjacent to brand new homes instead of overgrown vacant lots or crumbling warehouses). This is the core of the NIMBY problem in my view: building housing on my block has diffuse affordability benefits far beyond my block but very concentrated harms (living through construction, etc).
According to AAA (old data from three years ago) the cost to own a car in America is about $800/month, including depreciation, insurance, gas etc. So living in a community where you can take transit or a ride share easily can make up a significant portion of higher rent. Some. Not all. Just something to keep in mind in moderate priced regions.
Re: SF; a few years ago I saw a study where the researchers kept drilling down on why NIMBY's hated new development, and the biggest draw was hatred of already rich developers getting even more rich. So the green monster called jealousy may be a factor in a classism response.
The driver that both sides seem to be missing — or at least are not being explicit about — is that _density is necessary but not sufficient for valuable things beyond more housing_.
In particular, density allows for thick labor markets and increased specialization, making people more productive; and thick social markets, making life more enjoyable (try finding a date today to take to a show tomorrow in NYC vs. SF!). Those things take time for cities to develop and are dependent on things other than density, but they are valuable, and so folks will pay for them if they come to exist years down the line.
That's why:
- building more housing in NYC will immediately decrease rental prices
- building a huge amount of housing in rural China won't increase rental prices alone
- ...but building a huge city in rural China _and_ having an industry cluster form around it — such that people have productive work and enjoyable lives — _will_ increase rental prices
- inner city Detroit is both dense and extremely cheap, and building more housing won't change that
To believe that building more housing in NYC would increase rental prices in five years, we need to believe that more density makes living in NYC _even_ more valuable. Is that the case? I'm not sure — maybe beyond a certain point of density, it doesn't make us more productive or our lives more enjoyable. Or maybe there's no limit.
Regardless, the model of "Density Increases Local But Decreases Global Prices" is unambiguously wrong for places that can't build those other things — inner-city Detroit as an example.
>The two densest US cities, ie the cities with the greatest housing supply per square kilometer, are New York City and San Francisco. These are also the 1st and 3rd most expensive cities in the US.
Let's distinguish between the housing stock (quantity of homes) and the supply curve, which represents the quantity of homes as a function of price. Also note that price is determined by the intersection of the supply and demand curves.
>For example, if my home city of Oakland (population 500,000) became ten times denser, it would build 4.5 million new units and end up about as dense as Manhattan or London.
Are you thinking of a shift in the supply curve (moving along the demand curve), or a shift in the demand curve (moving along the supply curve)? If you shift supply, prices go down and quantity goes up; if you shift demand, prices go up and quantity goes up.
>Right now, more Americans prefer to live in big cities than there are housing units in big cities, so prices go up and these people can’t afford their dream. As new cities become “big” (by these people’s criteria), they’ll move to those cities, increasing demand.
Assume a two-sector supply and demand model, where we have graphs for Big cities and Small cities.
In the initial equilibrium, prices in Big cities are higher, because more people want to live there than there are homes. But note that Small city people still show up in the demand curve for Big cities: they're in the section below the equilibrium price, because their willingness-to-pay is low.
Now suppose that a Small city (somehow) transforms into a Big one. This shifts supply right in the Big city market, reducing price. If the price reduction is large enough, we move down the Big city demand curve so that some Small city people are now willing to pay the lower Big city market price. Note that Big city demand hasn't increased; it was always there.
(Exercise for the reader: what happens in the Small city market?)
At the micro-economic level, adding more housing units can indeed "lower" the median household price in a given metro area. Even in an affordable city like New Orleans while simultaneously increasing the adjacent cost of housing in certain neighborhoods via market stimulation. The best example I can give is an affordable housing project I worked on that was comprised of 79 affordable units for working artists (income qualified). It was renovating an abandoned school in the historically black neighborhood of Treme. The school had been left abandoned since Katrina (2005) and the immediate surrounding blocks reflected the slow decline of the nieghborhood hollowed out by dis-investment, subsequent broken-window crime and poverty.
Shortly after our project began construction, local builders and individuals began buying adjacent abandoned properties and either renovating/constructing new single and 2-family homes. By the time our project had been completed, the 79 units helped to lower the "overall" median rental price yet the adjacent neighborhood saw an increase in housing costs as gentrification took hold. This story repeats itself in other cities I've worked in like Denver, Washington D.C., Nashville, and Dallas. Phoenix was the outlier in that there's so much sprawl, gentrification can't really happen.
I have yet to see an in-depth study that quantifies the impacts of new housing units on market pricing at the overall median price level for a city and the immediate local affects of those same units on a particular neighborhood. I think the NIMBYs and YIMBYs talk past each other in this regard where each point to the relevant data that best supports their position without taking into account the complexity of how increased housing density affects are regional/metro and hyper-local. Which is why you have YIMBYs like Matt claiming more housing ultimately lowers the median housing price (nominally/briefly) while NIMBYs claim every new neighborhood development causes housing prices to rise and gentrification by yuppies/hipsters/bohos push out the the BIPOCLGBTQ+ artist/teacher/essential oil craftsbeing/zen gardner/retiree.
The conclusions you draw from your urban density & housing affordability correlation graphs are falling prey to the classic correlation is not causation problem. You acknowledge the potential reverse causation effect, but there are other reasons correlations don't imply causal relationships that you are missing from this analysis. In particular, you are not accounting for an omitted variable that affects both urban density and housing affordability: desire to live in e.g. Oakland vs. Manhattan, AKA housing demand. You correctly identify that induced demand is a thing that sometimes happens, but induced demand does not describe the whole dynamics of density and housing prices--you yourself say: “[People are] moving [to large cities] because they want to be in a big city - with friends, jobs, museums, and nightlife.” These are related but not perfectly correlated with density of housing for a multitude of reasons. Not all big cities are interchangeable.
Or: there are three relevant factors to the theoretical YIMBY argument about how marginally building more houses lowers housing prices, and you are kind of ignoring how demand/desire to live in a specific place is a separate thing from density, and should be controlled for when you graph a price function of density. This omitted variable biases your causal conclusions.
At least, that’s my read. If I got any of your arguments/thinking wrong, I do apologize.
“For example, if my home city of Oakland (population 500,000) became ten times denser, it would build 4.5 million new units and end up about as dense as Manhattan or London. But Manhattan and London have the highest house prices in their respective countries, primarily because of their density and the opportunities density provides. I don’t see why Oakland being able to tell a different story of how it reached Manhattan/London density levels (“it was because we were YIMBYs and deliberately cultivated density to lower prices”) would make the end result any different from the real Manhattan or London. But if becoming just as big as Manhattan or London would make Oakland more expensive, shouldn’t we assume that a little step in that direction would make it a little bit more expensive”
If Oakland all of a sudden had 10x the housing supply I would NOT think it would be as expensive as New York. I think you would have a ton of empty housing and princes would be dirt cheap as people tried to get something for their homes for which there was not enough demand...
The reason we don’t see his is that no developer is going to build a ton of supply beyond demand. Since it’s so hard to build this is rarely the problem. But we see it when cities de-populate. There was a time when Detroit homes in some areas were going for ~free. If someone added a bunch more supply in Detroit it would be unlikely to “induce more demand than supply”
It seems like a lot of the argument rests on “we can all agree it is true at the extremes, so doesn’t that counter the studies that show it is not true at the micro level”. But if someone (me!) doesn’t believe the extreme argument then the studies that show that induced demand does not outweigh increased supply effects seem totally plausible.
So I disagree with this argument for a lot of reasons.
I think the first (and easiest) counterargument is a money-on-the-table one. Let's suppose you're right; that if a town has a density similar to Athens, GA (where I live, good sized town, 100Kish people), then it should have housing prices similar to Athens (about 300K for a single-family home). Athens has lots of ordinances about what can and cannot be built, you are going to have to apply for permits, etc (this is true anywhere). So, just buy up 100 square miles of land in rural Georgia and build a similar sized amount of housing. If you did this, your profits should be astronomical, since you would have very little regulation to fight (this is especially true of building an Oakland sized city in not Oakland). But we don't see venture firms actually building brand new cities wholesale in America, so maybe this should tell us that there actually isn't money on the table here.
2: I think that this has been mentioned before, but jobs are probably a bigger driver of rent. I think this is especially true in single-employer cities. If you consider most rust-belt towns, their land value dropped like a rock not because more/less houses were built, but because the factories closed and were automated/went abroad (this happened to my hometown). You can even see this in bigger cities like Rochester, NY, which has a pretty high population density for a city, but not an extraordinary cost of living. It used to have Kodak as its #1 employer, but when Kodak closed, the city took a huge hit. Maybe I'm wrong about the numbers, I'm open to someone beating me back with hard data, but this is my overall impression that a lot of cities/towns lose land value from loss of employers, not housing issues.
3. I think that there's an important replacement rate issue. I do think that you're right in general, that as a city grows, housing becomes more expensive. But what I think is going to matter is how quickly housing grows relative to population; if a city could maintain 10% of its occupations as uninhabited, then I think it's land prices are going to be lower than a city with 1% uninhabited, even if their populations grow in lockstep. I think though that as a city gets bigger, its regulations are going to make it harder and harder to add new housing, and so the percent of uninhabited places is going to slowly drop, adding to the rent increases.
There is also a question of why any property owner would build so quickly that it forced prices down. There is an inter-temporal equilibrium here too as I explain in this paper.
Many people simply assume that changing planning rules will change how fast new building happens. Sure, it will probably change where it happens, but probably not how fast more housing is built across all substitute locations.
i think it would be helpful think in terms of elasticities rather than in levels. wine (nyc/sf) is more expensive than coffee (north dakota). but if demand for either increases, price responds very little because supply is inelastic. if more people want coffee, coffee producers just harvest a few more beans, hire a bit more labour, etc. marginal cost barely increases, and if the market is competitive same for the price. if housing supply were elastic, there would still be a gap between nyc/sf and north dakota (like the gap between wine/coffee), but it would be much less. the current difference reflects inherent factors (better amenities and higher real wages in nyc/sf) and the "distortion" of inelastic supply causing excess demand to drive a wedge between prices
the point about cities being in competition is just another way of saying demand is elastic (since cities are close substitutes for each other), so any increase in supply in a city won't lower own-city prices much. this isn't a problem if the people deciding on how many houses to build are "price-takers" (i.e., they don't consider the effect of their housing construction on overall prices). as long as MB > MC at current prices, they'll build and the market will clear. the current situation--where interest groups who *want* prices to rise use regulation to make supply inelastic--is the opposite. it's more like local political economy oligopsony than classical general equilibrium perfect competition.
i also like to think of it in terms of flows rather than stocks. suppose there are 1m houses in your city (stock). mostly people occupy their houses year after year. every year, 10k people show up in your city looking for a house. if there are fewer than 10k vacant houses V, they bid up prices. if there are more than 10k vacant houses, prices drop. it's the ratio of vacancies to people looking to search V/10k (what economists call "tightness") that determines the price (i.e., it's the 10k flow not the 1m stock)
if a city had a one-time extreme increase in its housing stock, but then implemented laws to make the future housing stock inelastic (i.e. the history of ny/sf), you would see low prices for a few decades, then over time as demand increases eventually prices would explode. this makes me optimistic because it means we don't have to make huge permanent investments in the housing stock, we just need to loosen things up enough so that every year, the number of houses coming onto the market exceeds the number of people looking for houses. this seems more manageable
Increasing supply has 2 effects (abstracting from the land price vs building cost distinction for now):
1) For any negatively sloping demand curve, it lowers prices.
2) Increasing density makes the city a more desirable place to live and increases wages through agglomeration effects. This shifts the demand curve for living in that city upwards/to the right, increasing prices
I think you're making the case that the positive correlation b/w density and prices tells you that effect 2) usually dominates and is what would likely dominate in Oakland if the Oakland were to build more housing.
I don't think this inference works because many amenities that make a city desirable aren't perfectly correlated with/entirely caused by higher density. That is, dense regions can be expensive for two reasons:
A) They happened to have nice amenities for reasons that are not caused by their density (e.g. nice beaches nearby). These nice amenities increase demand and prices.
B) Density causes better jobs and better amenities as in point 2) above.
A) would cause the correlation between density and prices to be positive (if weaker) even absent B).
For an example of A), were NYC density to reduce by 20%, Broadway wouldn't decrease in quality by 20%. Similarly, if Miami decreased it's population by 20%, I don't think it's amenities (the beaches) get 20% worse.
Similarly, if Oakland increased density by 20%, it becomes a moderately more desirable place to live because of B). However, if A) explains a substantial share of the variation in desirability, then Oakland doesn't attain the amenity/desirability/price level of a city that currently has 20% more density than it.
I think this means that it's totally plausible (though not guaranteed) that effect 1) is larger than effect 2) and that an increase in supply lowers prices. Indeed, this is what Matt's studies seem to show.
The missing link here is that by just focusing on housing prices, we are ignoring the other key components that go into what make people better off (what economists usually define as "utility"). People don't just isolate one component when making a decision, they consider the bundle as a whole.
Let's say that people's well-being has only two components: 1) quality of local amenities and 2) budget for non-housing consumption (think especially of tradable goods like iPhones and airfare).
It can make sense to move to NY from a lower cost city if you value the local amenities highly (walkability, Broadway), even if you keep the same income. You can consume fewer iPhones but you're still better off.
Now let's throw another important wrinkle in here: income. Big cities literally make some industries more productive (knowledge spillovers, scale), and some component of that productivity is captured by workers in higher wages. If moving to NY changes your after tax income from $50k to $100k, you can change your housing expense from 20% to 50% of your income and still have a larger budget for non-housing expenses ($40k to $50k) in this example.
The point is that narrowly focusing on minimizing housing prices is a dead end, because the other components of utility will tend to overwhelm it.
The YIMBY argument should be that construction lowers housing prices *holding everything else constant*, which increases utility - more people living in high productivity areas (more economic output for everyone) and having access to amenities that require density (if that's what they desire, and people wouldn't move to bigger cities voluntarily if they didn't want it).
If building more housing in a city would increase the value of real estate in that city, perhaps we can count on political forces in that city causing more such housing to be built. (Renters might not like it, but they probably don't have nearly as much influence over the housing approval process as land developers.) Consequently, cities that are not seeking to increase the quantity of housing are likely cities where building more housing would lower the price of housing.
There’s also the question of whether cities can actually generate a sharp increase in housing supply from the usual policies that YIMBYs advocate, such as mass upzoning. There are good reasons to be skeptical. Recent work by Cameron Murray has explained why “landbanking” is so common and why rates of supply in cities are closely tied to market demand conditions, not to zoning policies. You should check out his research:
There's a relatively clean theoretical answer, which is that demand is a function of social features (like good food, cultural events, ability to find your niche, variety of jobs...), and social features are a *superlinear function of population.* Social stuff is dependent on the number of possible pairwise interactions.. city X with population double that of Y will produce more than twice the number of patents, creative collaborations, etc... as well as more than twice the amount of crime, all things equal.
If social features scale superlinearly (and demand depends on it) and housing scales linearly, then you'll see housing prices go up as density increases. If you add housing without adding density, say by building outwards, you won't see the same sort of price increases.
But that's not all... what we should really look at is density and point to point travel time. A city that spreads out with good travel infrastructure may have housing prices similar to a city that builds up with mediocre travel infrastructure. I'd guess that population, how many other people you can reach within a fixed unit of time, and housing stock together predict most housing prices.
This analysis seems to have trouble explaining why anyone would ever build a detached suburban-style development in, say, Houston; all new construction everywhere should look as much like Kowloon Walled City as it is legally and architecturally possible to do so, because from the developer's perspective, density is a free money-making machine. In fact we basically never see anything like this, suggesting that short of maybe some Neom-like scheme on a truly grandiose scale, it is impossible for agglomeration-effect-induced-demand from normal building projects to come anywhere close to outweighing the supply increase.
What is the upper limit on geographically mobile yuppies ?
China is able to erect new ghost cities for a million+ people in a decade. If yuppies flood to a town, then there are no technological barriers to building at a pace of 100k/yr. I can't imagine 100k yuppies moving to Oakland every year. Inevitably, there will be a surplus and prices will go down.
The rise in local prices only persists when supply is limited. In fact, the local increase you speak off, is no different than scalpers re-selling new consoles at extortionate prices. This only lasts until manufacturers can get all their supply chains ready for the massive demand. Eventually, the manufacturers catch up, and the massive demand allows them to build their product at much lower costs due to scaling benefits.
With the deterioration of mid-western cities and their terrible weather, I'm surprised that internal migration to fair-weather states is as low as it is.
Everyone keeps wanting to bring up Tokyo, but they're ignoring the third biggest city in the US: Chicago. Chicago has world-class urban amenities and a broad range of high-paying employers. But housing prices are significantly lower there than in comparable US cities because they've consistently made it easier to build new housing.
Of course, you can bring up other reasons why Chicago might be less attractive, like Illinois politics and taxation, cold weather, crime, etc., but I don't think those are enough to explain the huge gap in housing costs. This kind of disproves Scott's model that there's a huge pool of people in the US who just want to live in a big city and growing bigger might attract them. If they were really as location-agnostic as Scott claims, they'd all be living in Chicago now. (Or maybe Dallas or Houston, the fourth and fifth largest metro areas in the country, that again have significantly lower housing costs).
So according to your hypothesis, homeowners should support construction of new housing nearby, because it will increase prices of the homes that they own. But typically we see the opposite. Are homeowners wrong?
Anyone interested in this topic should read the paper from Moretti and Hsieh on it. Examines the effects of reducing regulation for land use and how it’s related to employment, wages, and city size.
Maybe Houston and Austin are instructive here? Similar climate, similar amenities, similar density,, similar per capita GDP. The salient difference is land-use policies, and the rents in Austin are 50% higher. There's a lot of artificial scarcity baked in to housing prices in some places.
I'm tempted to ask if part of the issue is that the areas with more expensive homes relative to wages are simply areas with more established money.
On some level, in a classic economist "assume a can opener" sense, it's odd to have this difference between areas. If in some areas housing prices were higher relative to wages, you'd expect a movement in wages or in housing prices (or in people moving) to equalize it. Various reasons why that wouldn't happen ... maybe it's reduced costs in other parts of the budget (e.g. cars), maybe it's that they're more attractive places to live so people pay more for a better product, maybe something else.
Or maybe it's that in the areas with higher prices relative to wages, there's an added element of housing being bought with old wealth instead of wages.
In this view, some project city YIMBY utopia built out in the plains in Nebraska would be cheaper than NYC, simply because there's no rich trust fund babies.
I think the issue is that you are treating all types of building as the same, MattY is separating out hones and other stuff. People don't want to live somewhere because of the homes available (otherwise there wouldn't be lots of cheap property in the rust belt) but the other stuff. If you build more homes but keep the supply of everything else low (schools, shops, jobs, parks, etc) the prices will go down. If you build lots of nice new stuff but keep houses fixed price of housing goes up.
The big picture concept you're grasping is "spatial equilibrium", the idea that migration equalises quality of life. Quality of life has pull factors like "friends, jobs, museums, and nightlife" (agglomeration) and push factors like rents, travel times, and crowding disamenity. Basically, quality of life equals wages plus fun things minus costs and not-so-fun things, and this residual equalises across space. Imagine water levels equalising when the tide changes.
BTW there's two parts to "equalising". One is the direction: people migrate from worse to better. That's your "perpetual motion machine". Another is the impact: in-migration bids up rents, bids down wages, increases congestion, and worsens crowding. That's what stops the machine.
Example: Oakland gets cheaper and induces in-migration. More people = more bars = more fun, yay! In-migration continues. Bartender wages fall. Rents rise. Traffic gets slower. Bar fights increase. People stop arriving. Oakland is bigger. Maybe it's more expensive, or maybe it's cheaper if the growth was managed badly and had high amenity costs. Falling rents can represent falling demand just as much as they can represent rising supply.
This is spatial equilibrium, and it eats up housing price gains faster than we can induce housing suppliers to deliver them.
We totally need better write-ups of this concept. It's critical for zoning debates, and very underappreciated. Many people assume upzoning will change supply without changing demand. But that's wrong. (Some economists even embed this in models of housing density by assuming a "closed city")
Cameron Murray's stuff covers so many parts of this.
Doesn't this theory require induced demand to increase demand at >1:1?
Oakland has 500,000 housing units, plus 10,000 people looking for housing.
Oakland builds 1 housing unit.
Oakland has 500,001 housing units.
What happens next:
1. YIMBY: Oakland has 9,999 people looking for housing, so housing prices fall.
2. Scott: Two people hear that Oakland is a larger city with better night life and try to move to Oakland. There are 10,001 people looking for housing. Housing prices rise.
It seems like Scott's smoothly parsimonious pricing function requires that any increase in housing supply/population generates a >1:1 increase in demand. So if someone moves to an empty patch of land, that should induce two people to want to move there. Otherwise, constructing a second house on that land would cause housing prices to stay flat or fall.
I think you're saying "each marginal unit of housing built increases marginal demand more than it does supply", if I'm reading it correctly. I think this is obviously wrong -- the creation of 10 units of housing does not also create with it 10 people wanting to live there. If we think of it as a ratio of number of units demand created per unit built, the ratio is clearly not 0, but it's also obvious to me (with no evidence whatsoever) that it's less than one. Close to 1, perhaps, but not 1. The ratio also clearly depends on how big the city already is -- it's closer to 1 in NYC, and closer to 0 in the middle of the great plains.
There are also lagging factors here -- if you magic'ed 10 million houses in SF tomorrow, it'll take many years for the population to actually move there (after everybody already there all spread out to their own house). So the game is to build faster than people can move, otherwise house prices will go up as you build because you're not building enough.
Edit: I think your insight of the mobile young'un is broadly correct -- NYC is currently too expensive, so they don't move there. So the demand curve for these cities is actually very flat -- you can build 10% more and houses will barely budge. Another way of saying that house prices are inelastic to demand in big cities.
There is a problem with using median local income in this analysis. Restricting housing construction will generally raise rents which will price out the poorest residents, all things being equal (of course there are considerations like preferences around amenities and floor space and so on). This is part of why SF median income went up so much. Therefore building housing will lower measured median income as poorer residents move in, even though the incumbent residents are as rich as they were before (again ignoring effects on wages and so on). So you might expect housing construction to lower rent (a real benefit) and median income (a fake statistical artifact), thus keeping rent/median income where it was.
A good concrete example would be to think of someone working at Wells Fargo in SF 15 years ago. Even though rent/median income might not be that much higher in SF (growth of tech / low income people pushed out), for the typical incumbent SF resident, rent/income has probably gone up a lot, because banking wages (in this example) are determined nationally. (And then WF responds by moving jobs to NC to keep things in equilibrium). So construction is just this process in reverse - rent/median income will still constant but the typical resident will be better off.
I think what this misses is that demand is both number of people and ability/willingness to pay. The thing that NY has but other places on the eastern seaboard don’t is the global headquarters of the finance world. By connecting into that, the ability to pay goes up. Adding housing to the middle of Idaho doesn’t create a financial center and therefore doesn’t increase demand.
Not that this probably affects the central thesis here much, butfor any discussion of 'demand for housing' it's worth bearing in mind that the cost of housing is to an uncertain but significant degree artifically bouyed by its emergence as an apparently 'evergreen' investment asset. People buy urban property because the value is likely to rise, a prophecy which has become self-fulfilling over time.
I think you’re wrong because you seem to be misunderstanding the causation and some other factors at play.
Housing is not expensive in New York because it’s dense, but because that’s where the money and work to be done are. (Which began with the harbor, the river, etc.) That affects not just housing prices, but salaries, migration patterns, opportunities, culture, politics and regulations, square footage, demographics, etc. And yes, all that feeds on itself to increase demand which puts upward pressure on prices. But only because demand rises faster than supply – which is caused by people seeking money and work, not housing.
It won’t work if you start with the supply side. You can’t build a new New York, or 10x Oakland’s economy, by simply building enough housing.
People don’t move to places with housing, but to places with money and work (even if they can’t get a hold of it, as slums around the world illustrate). Ideally places with work and money would be the same places that have housing, but as so many dying towns, bereft of corner-stone industry, testify to, it’s not. Having the housing doesn’t guarantee the work.
Building more houses where there is high demand for housing *does* contribute to an already hot local economy, but has a net downward effect on housing prices.
PS: Note that new housing is always new, i.e. high quality, built for people with money, often in rapidly gentrifying neighborhoods. Newly built housing is rarely itself affordable housing (to local politicians’ chagrin). But it will make other, older housing, in adjacent or comparable areas relatively more affordable. Even if all prices go up, the prices for the old housing won’t increase as much as they would if there were less housing.
That also means that to understand housing prices for a city, you have to include the cost of housing for everyone who works in the city, even if they commute and don’t live inside city limits. (If public transportation is good, it’s not necessarily a bad thing to live in the suburbs or a nearby town.) We shouldn’t expect the average housing prices in SF proper or Manhattan to reflect the average rent/mortgage paid by the people who work there.
PPS: The difference between New York and Not-New York affects all the numbers: mean, median, modal, top and bottom percentile. I don’t think there’s any good number you can use to compare the cost and value of living in New York, to the cost and value of living on a plain in North Dakota or anywhere like it.
You’d get a better picture if you compare housing prices in different parts of New York State or NY metro area, which can be seen as part of the same market if you squint. And you don’t have to travel very far from Manhattan, before the housing market looks very different.
For good, generalizable understanding, it would probably be even better to use cities and regions that don’t have multiple world-class universities and industry epicenters, that attract people from the entire planet (and so doesn’t have practically infinite housing demand, like NY and SF). Looking at more normal cities will probably give a better picture of how supply and demand works in the housing market.
I was going to say this is a minor technicality, but actually the more I think about it, the more it may be a significant issue. In the early 2010's, the little town of Williston in the plains of North Dakota was actually *the most expensive* housing market in the United States:
It turns out that oil companies were willing to pay a lot for workers, and anyone who could have worked in construction could get more in oil, so construction got really expensive, and supply was not able to rise to meet demand. They probably developed *some* amenities that weren't there before, but the rise in price was really supply and demand, rather than the agglomeration effect for amenities.
The rest of Japan doesn't need good housing policy - it has declining population. Japan as a whole has declining population, but there's still some people that haven't moved to Tokyo, and as they do, Tokyo is still slightly growing (I believe).
Obviously, the time scale matters here. Certainly, if you simply disallowed anything more than single family dwellings you could prevent a sleepy town from becoming a major metropolis. So, certainly in the long term, there can be such a thing as induced demand (though whether this holds in expectation is unclear...not all possible cities will become cities). And what you are doing there is basically making the city less of an attractive place to live than it would have otherwise (maybe not to the people who want to live in a sleepy village but to the average person).
But, in the near term, we can model the amenities/jobs in the city as relatively constant. Or, to put the point differently, no one is deciding where to move by looking at the total number of units in that metro area. They look at the amenities and the price.
Thus, the only plausible way you draw more people to the city by building units is if house prices go down. In other words, absent the price signal what induces demand?
Now, in the longer term, more people draw more buisnesses, more restaurants etc etc... So certainly in the longer term you can induce demand but that happens after more people move in so on a longer time scale.
So yes, it's plausible that in the long term more housing leads to higher prices but only via reducing the prices on the margin in the short term.
Edward Glaeser (Triumph of the City) here would point out the relevance of the Z-axis (building height) and regulations preserving building stock, and in particular low building stock. San Francisco doesn’t have a lot of tall buildings. It has a lot of historic low buildings. So too does Manhattan, which brought us the elevator but also brought some of the preservation orientation that you are probably seeing in the data. Both cities could be denser and taller than they are. Enrico Moretti at Berkeley (where I teach) has a good paper on this that I will find and post a link to.
Japan has experienced a demographic transition, with an older population and few children, leading to negative population growth. There are whole villages where housing is very cheap since their owners have aged and their children are uninterested in the rural lifestyle. It's interesting that the problem is always "not enough houses (or water, or food, or jobs)" as opposed to "too many people".
Bottom line: more construction doesn't cause higher prices, ever, except in the sense that demolishing a 20-unit "average" apartment building and replacing it with a 20-unit "premium" apartment building will cause rents for that building to go up. There will also be a small effect on the immediate neighborhood, causing rents in nearby buildings to increase slightly due to the neighborhood being seen as more "premium". But housing prices in nearby neighborhoods will go down, because there is more supply of housing in the city.
You're wrong t think of housing prices as being driven primarily by housing density. Local desirability is far more important. In my part of Connecticut, house prices in low-density Avon are much higher than high-density Hartford or New Haven, because people with lots of money to spend on housing want to move to "desirable" communities - and one of the prime factors in making a community desirable is the quality of public schools. And public school "quality" (as measured by things like graduation rates, SAT scores, absenteeism, etc.) is influenced much more by the students attending the school than by any action the school teachers or administrators can take. So, house prices are high in rich towns because rich people want to live in the rich towns, so their children will associate with the children of other rich people, and absorb bourgeois expectations.
Re-consider your Oakland thought experiment: If Oakland were to build 10 times as much housing, why exactly would people with high housing budgets want to move in? They'd be moving into a hard-luck seaport city with poor schools and high crime. It wouldn't be for the awesome urban amenities, because Oakland doesn't have them. And no one will build them until the yuppies (or whoever has the money to demand the amenities) move in. Which won't happen because there's nothing to draw them in.
Extend your thought experiment to that wide-open plain in North Dakota: If you build a dense city with 3 million housing units to support a population of 5 million people, who exactly do you think will move in? Aside from the oil field workers. Would San Francisco tech workers move there as a more affordable alternative to San Francisco? If they want San Francisco's amenities, they won't move until your instant North Dakota city has the amenities. And who will build the bars and coffee houses the tech workers want if there are no tech workers for customers?
I think you have to separate out *housing* density from job and amenity density. The places you're describing have a lot of both. But they're not the same thing. Think even more micro-level and this is obvious. The most expensive real estate in any given area tends to be in a central business district with lots of jobs and amenities but not an especially large residential population.
If you increase housing without adding more jobs or more bars or more museums, then you're not inducing any more housing demand. You're just creating supply and prices will go down. There's a tendency of course for housing to induce amenities which induce housing demand and so on. But if you just build a bunch more housing (especially in an area that's already pretty saturated and doesn't have space for new amenity development), you're not setting up that circular dynamic. A lot of the big ticket amenities are fixed in the medium-run anyway. You can double the size of NYC, and they're still not going to build a second Met or bring in two more baseball teams.
There are some weird, off-diagonal places out there to confirm this. City of Industry, California has a population of 264 and a density of 17/square mile, which sounds rural. But it also has 67,000 jobs, and it's quite expensive there. On the other hand, the densest cities in the US technically speaking are several New Jersey suburbs of NYC, all of which are much cheaper than the city itself because they don't have much to draw people in. The slums of major cities in the developing world are some of the most densely populated places you'll find, but they've got amenities and they're very cheap (consider that the most densely large populated cities in the world are Manila, Dhaka, and Kathmandu).
If Oakland builds only so many new houses that all the related factors of attractiveness would be perceived the same (diversity and density of jobs , higher quantity and quality of cultural/ social/ medical infrastructure etc.) I believe *this one* specific activity would lead prices to go down or at least to stagnate. The problem might be, that if Oakland builds only so many houses & the level of attractiveness stays exactly the same, the demand-effect of people wanting to move to Oakland is already surpassing the effect of the additional houses. So in effect prices go up. However, they do not go up as much as they would have without the additional homes. So Oakland, even if they were the only ones to build new homes, in this scenario is better off than if they didn't. (Except if they are perceived to be the only place around with a great YIMBY culture and sensible politicians, which might also increase attractiveness. But then it's *not* purely the number of homes again.)
I believe there are thresholds, and the attractiveness of our exemplary place O. rises only if there are so many more people, that this attracts, say, a new university, two major industries, a new high quality theatre ... or at least lots of new restaurants and cafes and local arts culture (I don't know Oakland, so using examples from my area). In other words, if it is perceived to be a bigger *and* therefore more high-quality / more opportunities / hipper/ ... place.
I would be interested to see good data, how density correlates with housing prices in my country. And how city size correlates with the latter. Or what about European cities ... okay, that's maybe a bit unfair.
And yes, I think building new homes makes sense, but I think housing prices need to be tackled by other means.
>Or doesn’t it (as Yglesias argues) allow an economic perpetual motion machine, where you just keep building houses and generate infinity money as the price of each keeps going up?
Apropos of nothing, the canonical Urban Land Rent model in Econ, Alonso-Muth-Mills, assumes all workers labor at the platonic center of the city, an infinitely tall skyscraper which uses no land. I often think of this building and its employees.
Here's why you're wrong in a single sentence: Demand causes high prices, not new units.
Prices are high in SF and NYC because those are desirable places to live for a huge number of people. People all over the country and the world would live there if they could, and prices reflect that. The fact that the densest cities are the most expensive is true. But the high prices are not caused by density - rather, the density and the high prices are both a consequence of crushingly high demand.
I generally agree with your hypothesis. More housing is more supply which should reduce prices. But a simplistic Georgist analysis would seem say that more housing increases the value of all the other property around it, which includes other housing. It doesn't seem impossible that in certain circumstances, the 2nd effect could outweigh the 1st.
I'm curious if any Georgists would like to weigh in on this one?
You vastly overestimate the percentage of Americans that want to live in a dense city. Some Americans do, but most don’t, so density is at best a wash and probably reduces demand. Thus, the causation flows almost entirely in the other direction, with high demand causing high prices which cause density.
When considering the price of a thing, demand is an important variable that can not be reduced to supply alone.
There are volumes of ink that can and have been spilled about the details, but you're not getting this to reduce to a function of one variable.
The argument is that, on the margin, increasing supply will reduce price of a particular good. An observation that the price of that good is currently highest where the total supply is highest doesn't really engage with the argument. Second-order effects are table stakes.
What about the following theory, which is consistent with both your arguments and the studies: if you build houses, prices will temporarily decrease, but then they will increase later after some amount of time (e.g. 20 years).
The delay in the price increase comes from the long time it takes for density effects to materialize: people don't just want to move to a dense city, they want to move to a city with jobs. However, it takes employers time to set up in a new city, and therefore, the increase in employment opportunities that follows from density might take time.
(As an aside, if people want to move to dense cities due to jobs and if those jobs materialize disproportionally in dense cities, then it is a bit of an economic perpetual motion machine; the efficiency gains from density must be *huge* to justify such a story.)
I found this a bit upside down and tried to put down some notes. Not ready yet, but as some points seem to recur in discussion ....
First, what if what if density doesn't drive prices, and prices do not drive density? What if instead, there is a third underlying variable that affects both. That variable might be attractiveness. Okay, the text mentions this.
Second, but what concretely is making a place or city attractive? Elements might be: the status of the city as (local) capital, the natural environment in the surroundings, the city's image as being hip or you name it, the amount of available jobs and career prospects, the quality of cultural infrastructure (theatre or whatever you are into), the perceived likelihood to find like-minded people, etc.
Third, what then is the relationship between attractiveness and density? I believe beyond a certain threshold density as such is not attractive to (most) people. I believe desity is an unperfect proxy for some of those factors of attractiveness (availability of high quality theaters) - and unrelated to others (beauty of the natural surroundings).
Forth, instead of calling it a proxy, one could also describe a feedback loop that involves density somewhere: more people moving to a place leads to higher density, this leads to more industry being attracted and more needs for culture, which leads to more jobs & more new theatres set up, which leads to the place being more attractive, ...
Fifth: why is this important: Because it impacts the hypothesis that more housing leads to more density leads to higher prices. Whenever more density doesn't lead to 'higher *perceived* level of attrativeness', then in this model, it also shouldn't increase prices.
Sixth: But what about the feedback loop? Well, it is probably happening in many areas, but there need to be other factors involved, and it doesn't need to happen always. Among others see next:
Seventh: There are probably also thresholds. Not every 1% of new housing leads to more attractiveness factors or the city being perceived as 'bigger'.
Eigth: Isn't city size the factor and not density? I suppose only few people really want to live in a dense place as such. Maybe city size is a better predictor for those attractiveness factors? Would be interested to see if prices are correlated better or worse.
Ninth: Correlation is not causation. That's clear to everybody here I suppose, but I had been missing a more explicit follow-through on this in the text.
Tenth: Why *perceived* level of attractiveness? It's all in the eye of the beholder.
TL;DR More housing lowers rents immediately. More density slowly increases rents.
I think it's important to be a bit more careful about the timing/delay of impact when thinking about the different steps in the causal diagram here.
Here's my mental model:
A) More newly built housing -> People in already in city consider moving to that newly built housing
B) Rent prices come down (on quality adjusted basis) -> People in other cities consider moving here
C) More people living in the city -> demand for the city goes up, since some people like living in denser places with more people and more businesses
D) Many high income people -> new businesses open that cater to high income preferences
E) Many high skill people living in a place -> more high paying employers set up offices in city
F) Many high paying jobs -> people want to live in the city (housing demand goes up)
In terms of how quickly these channels operate. I think A>B>C>D>E>F
Channel A is the fastest to act. So new housing very immediately affects the set of options of current residents and starts to lower rental prices (relative to what they would have been under the status quo).
This is partially offset by B, with people seeing the lower prices and considering moving to the city.
Channel B can at most partially offset the impact of A (since by definition B is people who are moving *because* the rental prices are lower). Considering moving to a new city is a slower process than moving within a city and as a result, this channel also generally reacts more slowly than A.
Channels C + D are related to the impact of the changing population on the desirability of the city (either directly through the influence of peers or via new businesses that open).
Both of these are even slower processes. It takes time for the population to expand in response to the new housing and for people inside/outside of the city to notice the changing demographics. Setting up a new business is also a bigger investment that takes more time compared to a family moving between apartments. Also notice that these channels only make sense if the population is growing, but if B is growth caused by lower prices, then it's still hard to see how this could lead to a net increase in rental prices (relative to status quo).
Channel E (employers choosing where to locate offices) is much much slower. Many cities wish they could convince more high paying employers to move to their city, but it takes a lot to overcome the inertia. Also high rents for housing and office space directly push against the desirability of a city to employers, since those add to the cost of moving to the city.
Channel F (people moving to areas with good jobs) doesn't happen until E happens, so it is yet more delay in the response.
In my view, the new housing lowers prices* in less than a year after the new housing. All of the other factors kick in much more slowly over a period of years, and at most partially offset the decrease in rents.
* on a quality adjusted basis, relative to the status quo of not building.
Not understanding why you are only talking about reverse causality as a possible challenge to your interpretation. What about omitted variable bias: there is some other underlying factor that is associated with density, which is in fact driving higher housing prices. I would tend to think that OMB is a huge problem here, as there are many very dense slums in the world that are not in fact not expensive at all, most famously Kowloon City.
I have a few guesses as to the origin of OMB, most importantly, jobs. There are certain locations that are just much better in terms of jobs. If you want a job in the media or finance, you should probably move to New York. If you want a job in government, you're going to have to move to DC. If you want a job in the movie business, LA is really your only choice.
If jobs are in fact driving much of the increase in housing demand, building extra housing is not going to raise housing prices. Moreover, if these are relatively high paying jobs, then people are still going to come even if you don't build anything, because they'll be able to afford more than the average resident in the area. And this is exactly what has happened in the bay area: the working and middle class has steadily been pushed out by tech professionals.
This seems like a far more natural interpretation than people move to an area simply because there are a ton of other people already in that location. New York's density is a strike against it in my book: too noisy, too crowded, too stinky.
My guess is that what you're really thinking about is agglomeration effects, but returns to agglomeration come from industrial agglomeration, not from residential agglomeration. It's a critical mass of businesses in a given location that makes it into a jobs hub and attracts migrants.
I think this this analysis suffers from three problems:
(1) Reducing a complex multiple variable and step interaction to two factors (housing density & cost). There are a large number of motivation factors which you touch on but then discard from your model to use density as a proxy(Jobs and income being the most significant).
(2) confusing a correlative measure (density) for a cause. Density correlates with cost because they are both result from the same cause (Jobs and income). If stripped of its jobs, Manhattan would have low housing cost, despite the high housing density. Conversely, If you built 5 million units in Oakland or the North Dakota, the cost plummet, at least initially. This transient phenomenon only makes sense if you include the omitted variables.
(3) Explaining examples of urban population decline. There are many cities which have experienced population decline due to loss of industry. If population density were the primary motivational driver, this would not be the case.
At best, it can be said that people follow jobs, Jobs & employers follow population density, and higher housing density and housing cost both *result* from competition for increasingly concentrated locations.
If jobs opportunities are held equal, people will flow from high cost to low, not the other way around.
I am skeptical of the two charts you showed at the top because: (1) They use survey data when there's better data available. (2) The category "Urban Area Corresponding to Major Metro" seems like it allows the authors some choice in what to include or not include. The second one also seems a bit arbitrary - including San Diego & Cleveland, but not Minneapolis. (3) Neither of these includes "Podunk" or "Empty Space", so connecting them to the hand-drawn chart is extrapolation.
Is it better to use housing price or housing price divided by income as the vertical axis? I could see arguments in both directions. Both housing price and income probably change as more housing is built.
I am also confused by the relative positions of Miami and San Francisco in the two charts. Was Miami significantly more expensive than San Francisco in 2015 and significantly less expensive in 2019? If no, then the error bars on the data are large enough for me to not trust the trendline. If yes, then maybe someone should figure out how to do what Miami did.
Some initial observations from looking for the highest things on the list: (1) The most expensive core-based statistical area is Vineyard Haven, MA, a little island off the east coast. Manhattan might be even more expensive - it's not listed separately - but Martha's Vineyard is more than twice as expensive as metro New York City. (2) San Jose, San Francisco, Los Angeles, and San Diego are all in the top 25, but so are Santa Maria, Santa Cruz, Napa, Salinas, San Luis Obispo, and Santa Rosa. It seems like California (or at least coastal CA) is unusually expensive, even for small cities. (3) Hawaii and Rocky Mountain ski towns (Jackson, WY, Breckenridge, CO, etc) are also unusually expensive.
My guess is that there is still somewhat of a positive trend between housing price and city size (or density), but that it is noisier than these graphs show, especially for smaller cities. It should be possible for a city which is currently surprisingly expensive relative to its size to shift down and to the right by building new housing. For example, on the top graph, if San Diego became like Las Vegas.
I agree that Tokyo should not be placed on these graphs. But it would be interesting to see where Tokyo lies on a graph with other Japanese cities. In a similar graph for Japan, is the slope positive or negative?
I can't follow your logic. Primarily cities are composed of a few moving parts. Primarily there is a source of a harbor or natural resource that fuels primary demand. Think of the ports in Chicago (and the stockyards during the turn of the century), or other major port cities. Indianapolis' main driver of economic growth is its interconnection of highways for example. These are elements of any city that can't be changed or relocated. The Chicago mercantile exchange is in Chicago because Chicago outbid St. Louis for stockyards (fascinatingly because they put in a sewer system) and because they have access to a river and ports. Surprisingly, not many people know that Cleveland, Ohio used to be much bigger than Chicago. This maps onto countries as well. Countries that are farther away from ports tend to be poorer, while countries that have large amounts of natural resources and are historically non-European tend to be resource cursed to exploitation.
If it were cheaper and more efficient (ie made more money) to have the Chicago Mercantile Exchange in Pidgeonforge TN it would be in Pidgeonforge TN. But it isn't.
So, cities that have a primary resource that can't be relocated tend to have a primary driver of economic growth that is stable through economic fluctuations. Software development can done anywhere, and often is, but the port structure of a Seattle or San Francisco won't change.
As to the cost driver of housing primarily? House prices in the long term appreciate in those places where costs are expected to go up. And prices are more likely to go up in those cities, over a ten, twenty or thirty year term, that are near resources that are structurally economically self sufficient. Were there some way in which Topeka Kansas had a structural comparative advantage over a port city on the coast then people would all move to Topeka Kansas and crowd around that source of economic development.
The farther you drive away from a major port the less expensive housing is, typically. Meanwhile ports tend to be geographically landlocked or have limiting geography, such as the LA basin, New York, or San Francisco.
Port cities are comparatively expensive because the ports are continual sustainable engines of economic growth that don't migrate with people with the fluctuations of the economy. That's it. There's not a whole lot of mystery here. This prisoner's dilemma unstable state Nash equilibrium notch is largely incorrect.
The NIMBY problem makes housing prices worse in those areas where people take over zoning commissions, but in cities that have a port housing in the long term will remain expensive. In cities without a port NIMBYism is much like price controls - all it will do is disincentivize housing even further and lead to urban blight. The less affect a port (and it's subsequent markets) has on sustainable long term effect on urban growth the worse NIMBYism will be for the economy because during a downturn prices will remain high. This is why NIMBYism in San Francisco is comparatively worse than say, NIMBYism in New York or Chicago as its economy is more heavily dependent on the tech sector which is highly cyclical. High prices during a downturn lead to people moving away, lead to empty store fronts and then urban blight which becomes self reinforcing.
You're overthinking it.
EDIT - On further review, I can't tell if you're joking or not. You could make everyone richer by burning money and causing a small amount of deflation, but I wouldn't recommend it. This looks like idiot bait.
I think what this argument is trying to show is, if Oakland got 10 times bigger, *with the same restrictions on housing development*, then the price of housing would go up. Which I think is correct, but it's controlling for the wrong thing, because loosening restrictions on housing development is precisely the mechanism that YIMBYs want to get more housing from. (One question about the scenario in the post is why people would build more housing if it's not easier to do so.)
What YIMBYs are saying is that reducing restrictions on housing leads to more housing and lower prices. This is clear from thinking about the supply and demand curves. In the absence of artificial restrictions on housing development, we would have a nice supply curve that goes from bottom left to top right. With restrictions, developers can't build as much as they would like, and basically the curve hits a wall at some amount of housing units. So the equilibrium with restrictions occurs at a lower quantity and higher price than the equilibrium without the restrictions.
The 1st order effect of loosening the restrictions is to move the situation closer to the equilibrium without the restrictions. If you read "The Economic Implications of Housing Supply" by Glaeser and Gyurko (2018), you see that the housing cost doesn't go too far above the minimum cost to build, if your restrictions aren't too bad and the housing market can adjust to demand (their example is Atlanta). For the Bay Area, the price is more than double the minimum production cost (these are both shown in their Figure 2). So there's a lot of room for prices to go down in places like the Bay Area.
What you're pointing out here is that there's a 2nd order effect, where having a bigger and denser city makes the demand go up. This is true, but I suspect that the effect is just smaller. "Growth, innovation, scaling, and the pace of life in cities" by Bettencourt et al. (2007) shows that if you double the size of your city, a lot of metrics around socioeconomic efficiency of the city goes up by ~15%. So things like your GDP per capita (and crime rates) grow by ~15% if you double the population, which explains why people want to move to big cities, and would make it more expensive to build housing (higher labor cost) and make people willing to pay more for housing, but it's not big enough to counteract the 1st order effect.
There's probably better studies to support what I'm saying, but these are just things I can think of off the top of my head.
You're working with a theory that there is a pool of people who want to live in big cities, but individual big cities are to some extent substitutes for each other. This pool can be overwhelmed with sufficient supply, but if it is not overwhelmed then big cities will be expensive *because* they are big - there is unsatisfied demand for dense living.
I disagree. I think people want jobs, and they move to big cities because this is where the jobs are.
"But that's just adding another step! The jobs are in the big cities because that's where the other people are!"
To an extent this is true. You can set up a sandwich shop in Sydney and sell a lot more sandwiches than if you set up a sandwich shop in the Great Sandy Desert.
BUT it isn't enough! Each additional person creates LESS THAN one extra market-rate job. If this was not the case and density fed on itself in an ever-increasing way, we would see cities grow ever bigger and bigger, with the goliaths outcompeting the giants. Why do some people live in SF rather than New York City? It's not because SF is a discounted option relative to the more populous and therefore more desirable NYC.
As a further counterpoint, consider declining cities like Detroit. If density feeds on itself, and it is the mere fact of more people creating more demand, then these cities should have not depopulated. Detroit was bigger than many other cities that have not shrunk so it's not simply a scaled-up example of the broader urban immigration phenomenon.
What you're missing is that a city needs an underlying economic engine. This is very obvious in small towns where there is often one dominant local industry, and the whole town's fortunes rise and fall with it. Not everyone in Nangwarry works in the forestry industry, but it's the health of the forestry industry that decides what housing demand is like and how many sandwiches you can sell there.
This continues to be true even for mega cities like NYC or Tokyo. Yes, they have so many people that there are lots of jobs supporting all the other people there, doing things like cleaning, car repairs, etc. Yes, there are so many people that they are good places to set up industries that could be based anywhere but want a network of similarly-skilled people in close vicinity. But there is a fundamental driver of why these places are big - they have a geographically fixed multiplier of economic value that is extremely powerful. That engine is a port.
There is a reason that all of the world's biggest cities are located on coastlines, and it's the reason that Adam Smith identified in The Wealth of Nations. Sea trade is extremely efficient. If you have a port, you can cheaply trade with every other city that has a port. This in turn allows you to unlock greater division of labour and greater efficiency than would otherwise be possible.
That's not the whole story of course - you need to have things to trade through the port. But it's a key reason why everyone in SF and NY doesn't just move to somewhere like Austin where they're cool with building stuff and they can have a super-dense mega city.
I think D.'s and Robert Donnelly's comments together explain it well.
You're missing an intuitive understanding of causality that maps back to a logical framework. I recommend Judea Pearl's Structural Causal Models, which he explains simply in his Book of Why. Note that the Book of Why is imprecise in a few parts, but it's a good enough start. The "correct" explanation is given in "The Do-Calculus Revisited," also by Judea Pearl, though it's a harder read.
I want to address 2 points, the long term correlations vs short term studies and the potential gains for a defector city in scotts hypothetical.
For long term correlations, recall that its possible to be a failed city. Baltimore was the 2nd largest US city in 1840, now its 30, and not just because a bunch of western cities overtook it, many eastern ones did to. In regards to induced demand, if a city is a long term success, it will have many houses and high prices ie NYC. Within some margin of acceptable housing policy it could be expensive relative to its success or cheap, but still bound to that curve. The studies show how a city can overperfom its position on the curve. But it would be foolish to think you could easily control a citys overall success. A nimby policy on the premise of capping housing prices by capping city success is likely to become baltimore cheap, not upper west side 30
years ago cheap.
Success is relative. The defecting city would see short term increase in housing prices as it under perfoms its spot on the curve. Long term will be driven by success but intentionally restrictive policy is likely to cause the bad kind of price drop.
Suburbs dynamic is different because its success is largely a function of the nearby city thus exogenous. It can have no housing growth but still be very successful and therefore expensive.
I see statewide control a la CA as not primarily resolving a prisoners dilemna between big cities, but forcing suburbs to upzone along with the successful city they are attached to.
Finally, success is also good. The YIMBY appeal isnt just globally lower prices, you get 1) more successful city 2) overperform your spot on the curve. Win win.
how do we account for slums? clearly it's possible to build housing at very low cost in very high density locations, in the absence of any regulations.
If you believe that, at the margin, building a thousand new units of housing in Oakland would cause housing prices to rise because of the induced demand, then you must logically believe that at the margin, tearing down a thousand units of housing would cause housing prices to *decrease*. Do you?
I'll even throw in the explanation of why it "should": some of the people living in the houses that were just torn down will say "screw this, I'm moving back to Stockton", and some of the people on the outside planning to move to Oakland will say "meh, the city doesn't seem so hip and cool now that they tore down those apartments, so now I'm rethinking this move". Those are things that will absolutely happen. They are the mirror image of the things that will happen if new housing is built, that lead you to believe that building new housing locally increases rents.
I'm pretty sure you recognize that this is false - there will be *some* people who abandon their plans to live in Oakland if you tear down a thousand apartments, but there pretty obviously won't be so many of them as to drive down rents at the margin. And at the margin, the difference between -1000 units and 0 units, is going to be very similar to the difference between 0 units and +1000 units.
Alternately, I look forward to your argument that Oakland should help its fiscally stressed renters and can't-quite-afford-an-apartment homeless by tearing down homes.
I think the essential chain of causation is something like:
1. Faster economic growth causes higher prices
2. Higher prices cause higher density (and other negative externalities)
3. Higher density (et al.) causes slower economic growth
4. Slower economic growth causes lower prices
So you have a dynamic equilibrium. Density (and other negative externalities) is why economic growth and prices in clusters like NYC or Silicon Valley don't just explode upward endlessly. Density is a brake on growth and prices, not an accelerator. It is correlated with higher prices because higher density is needed to fully brake the growth of a region with greater economic potential, but the density doesn't cause higher prices but the opposite
> Wouldn’t the alternative be some kind of highly unparsimonious pricing function like this?
I don't get the argument of this graph. Doesn't your imaginary graph look like you're talking e.g. about Minneapolis-St Paul (on the real graph) and just pretend that all other cities fall exactly on the line?
> This is a coordination problem: if every city upzones together, they can all get lower house prices, but each city can minimize its own prices by refusing to cooperate and hoping everyone else does the hard work.
Well, you can also spin the opposite narrative just as easily!
Assume cities like high land prices (eg because it gives them more tax revenue, if they use property taxes or land value taxes). Then each individual city can get more revenue by building more, but overall all of them together might perhaps not make more money.
Building more would be a 'tragedy of commons' in this situation.
---
Overall, I think your analysis is more-or-less correct, but that property taxes (or even better land value taxes) can fix the incentives for cities, so they can build up without having to coordinate with other cities.
Yes and no... the book I would point you to is Alain Bertaud's "Order Without Design" if you want a good, reasonably accessible foundation in how the economics of cities work.
Essentially a city is like a big party - it's a huge market that becomes exponentially more productive the larger it grows... so long as the city can effectively mitigate the downsides of proximity and density and avoid having the party fall apart after a few crucial guests leave. The biggest issue is maintaining mobility - most US cities are car-dependent. Car dependency allows cities to sprawl out and provide a lot of living space per person, but it also makes moving around very difficult above certain densities - thus making it hard to build an effective city of more than a few million. To get bigger, cities have to get more creative with their transportation networks.
New York and San Francisco are big, very successful parties that everyone is trying to get into. In one sense, they are unaffordable because it's very hard to get into them and allowing more people in would make them more affordable. But in another sense, it is the fact that so many productive and attractive people are clustered together in these cities that makes them so attractive and hence unaffordable.
It is certainly not true that building more housing makes prices go up. But to strong man this argument: ultra high prices are a drag on growth in very productive and attractive cities. Building more housing will lower prices, which will in turn allow more people to live in the city, further increasing its productivity and in turn attracting even more people, causing prices to eventually go up. Sort of like if you start turning away guests at your party just because you don't want to have a big party. Turning away people doesn't make your party more accessible (it makes it more exclusive obviously), but maybe it will stop the party from really gearing up and becoming even more popular.
So the real conflict is over urbanism: do you think it's a good thing for lots of people to be able to live together in cities and benefit from them? As a city planner and urbanist, my answer is yes: I think cities provide all kinds of benefits and I defer to people's expressed desire to live in cities. Outside of Manhattan, I would argue most US cities aren't very dense at all and that there's no reason to be afraid of letting them densify and shift towards different forms of transportation. Right now I think we have a chronic undersupply of the kinds of urban places people want to live in and that are capable of providing satisfying and high-paying job opportunities and life experiences in an advanced economy, so I support letting people build more cities.
He cites plenty of econ papers, going so far as to explain why the ones that concluded the opposite were mistaken or what they failed to include. I think you can start with his post and its links, and go as deep as you like to satisfy yourself about how it all works.
The key is that no city is a closed system: people move to different neighborhoods within a metro area, and new arrivals to the metro area have a choice as to which neighborhood they go to. Building new housing in one of them causes average rents to go up there because (A) you just built new housing, which is probably priced above median itself, and (B) the signaling that this is a good yuppie neighborhood causes follow-on moving. However, that construction and those moves *prevent* gentrification elsewhere in the metro area, because the people moving to the desirable city are able to buy new-construction housing rather than apartments in an older locals neighborhood. And the yuppies don't *really* want a 100-year-old apartment anyway (they'd rather the new construction), they'll just take it if that's all that's available in the neighborhood they want to live in, and they can afford to outbid others.
Are new builds like old builds? What's the elasticity of demand?
For example, in Pittsburgh's recent revitalization many of the new apartments are swankier and more upscale than lots of the old construction. A good deal of the renovation notably increases amenities. So yes there's more housing capacity but all the new capacity is more expensive for real reasons. So if you build new housing that's more expensive than average and people move in, you can raise the local price even as the movement of people decreases demand elsewhere. (And in fact, this is exactly what you see in Pittsburgh. Sure the new stuff is cheaper than LA/SF/NYC emigrants are used to, but now all the other prices have room and pressure to grow.)
If you build housing that resembles an average or a typical spread, you'd imagine it would decrease or at least not increase the price. You could make an elasticity argument a la Malthus, and argue that the price will fall, then some new marginal people might move in and the price would stabilized where it was previously because there was some reason it was that price in the first place.
To expect the local price to go down purely on the weight of new builds, you'd have to build so much housing that you overwhelm the elasticity of demand. In a place like Pittsburgh whose big appeal is "a cheaper silicon valley", there's probably significant elasticity! The cheaper you make it, the better that appeal! That could be an unrealistic amount of housing! If your goal is to lower prices you're better off making a structural change like WFH or public transit.
I feel like there should be a solution here where we let new development pay off renters in the area. For instance, maybe we give renters in any zone that approves new building in a year a share in future increases in overall property tax. It probably wouldn't have to be a very large amount before it convinced them it was worth allowing building (even if they may have to move away they keep that income stream).
Alternative hypothesis: New York City and San Francisco (and maybe also LA) are weird exceptions because they control the commanding heights of the culture, and therefore are viewed as uniquely desirable places to live. They are desirable not because they are high density, but because they control the News Media / Social Media / Entertainment Industry respectively, and thus occupy a disproportionate amount of mindspace, making a lot of people willing to overpay to live in them.
Chicago / Houston etc don't occupy this kind of privileged position in the culture, and see much lower housing prices.
How does building attracts more people to the place if the marginal effect is raising housing costs?
Like, if your city has a market settled at 100 housings going for $100, that means at that price there's only 100 people that want housing there. If there were more than 100 people willing to pay $100, the price would go up until the market cleared.
If you build 1 housing and that makes the price go up to $101, now there's less people that want to live there than before! Building houses would make cities shrink in population.
If you built 10 houses on a North Dakotan plain and no one bought them, wouldn’t the price go down?….until such time that at least 1 or 2 people bought in…at which point the smallest sight of demand might at least plateau prices at a new (but below initial) level?
It seems to depend on (at least) an ill-defined balance of land availability vs scarcity, location desirability, and hence the result supply vs demand, to even begin to speculate on an answer. My suspicion is the relationship will not be linear. Adding 8 million people in Oakland doesn’t automatically bring with it Manhattan prices, unless you have with it manhattan amenities, and Manhattan geography.
Hold on; if this argument holds it is _exactly_ an argument against YIMBYism: YIMBYs harm themselves and those around them. (In housing prices. Maybe the better quality is worth it.)
Yes, it's better if everyone agrees to enforce YIMBYism globally in this model, but that's not how it works, and it does mean that people who oppose YIMBYism in their own location are acting economically rationally.
I think what's missing here is the type of housing being built: is it single family or apartment AND is that dense apartment market housing? Non-market housing isn't affected by the market as much and can be cheaper. It's also not profitable so isn't common.
A possible counterexample here is Seattle. Like SF, Seattle's seen a massive amount of immigration over the last few decades, to a large extent of high-earning tech workers (in fact, proportionally more people have moved to Seattle than moved to SF in the last 30 years, both to the city itself & to the metro area). However, unlike SF, Seattle & the surrounding area also allowed a ton of construction, nearly all of which involved/involves large density increases on already-occupied land. The near-complete reconstruction of South Lake Union is the most striking instance, but you can see this happening all over the city (f.e. the ongoing transformations of Roosevelt & Northgate resulting from the opening of the most recent Link extension). Notably, Seattle's median rent also hasn't risen nearly as much; said median rent is far closer to the national average (possibly even slightly below at the moment, if Zillow's rent numbers are to be believed).
One important thing to note here is that Seattle's recent increases in housing supply have been mostly driven by large local increases in density. In some ways, the area faces the same kinds of geographical problems SF does (if not to the same extent), being quite hilly & hemmed in by water. In this, it's unlike the sunbelt cities that frequently get cited as evidence for "building housing to accommodate immigration keeps housing costs down", since there's little in the way of land to build on that doesn't already have something on it.
Possibly contributing is that the upzoning in the area hasn't been limited to one municipality. Several of Seattle's suburbs have also gotten in on the act, with cities like Redmond & Bellevue also seeing enough apartment-type housing construction to radically reshape (or, in the former case, practically create from whole cloth) their downtown cores. To the point about this being a coordination problem, perhaps part of Seattle's success is that the city & its suburbs managed to at least somewhat coordinate here.
Think of China's ghost cities / apartment blocks. Prices surely can't be that high there. Maybe the answer is that developers are good at their job, and build supply where theres demand for it?
English is the language with the most worldwide speakers. English is also the language with the most second-language teachers.
Does that mean that adding more teachers is an important way to increase the attractiveness of a second language?
Maybe, in extreme cases, but not really.
Convenient teaching is helpful, but not decisive, for people choosing their second language. Plentiful city-center housing is helpful, but not decisive, for people choosing their metro area.
In both cases the teaching/housing is a lagging variable that responds to, but only at less than 1-for-1 contributes to, the underlying desirability.
(BTW, Scott, I keep feeling that I haven't quite grasped your actual uncertainty, and I'm wasting both our time on things you already get. That's frustrating for me, and I bet for you too.
Maybe you can restate your thoughts more precisely in a week, after you've read everyone's responses?)
The problem is price and quantity are jointly determined by demand and supply. The exercise that Yglesias is proposing is akin to saying “imagine supply increases then price must decrease to a new equilibrium where more quantity (housing) is traded.”
This is nothing more than the standard Econ101 law of demand applied to housing markets.
The correlations that you show are a big no-no in econometrics. You don’t just regress quantity on prices (or viceversa) and give that anything close to a causal interpretation. Why? Because demand and supply jointly determine prices and quantities. Large traded quantities might be associated with high prices if demand is high relative to supply, low prices if supply is high relative to demand, or anything in between.
The most reasonable explanation for the correlation we see is that places in where supply is large demand extra large (NYC), so prices are high. In contrast, in places with low supply of housing demand is non-existent (North Dakota countryside), and therefore prices are be low. If you run a regression of prices on quantities you’ll get a positive coefficient, but the effect has nothing to do with quantities and prices and everything to do with the fact that in places where supply is high demand is even higher, and in places with low supply demand is close to zero.
If we go meta and ask why is demand so much larger than supply in NYC, this argument naturally ties to YIMBYism. The reason is that supply is artificially constrained in NYC through red tape and regulation. Removing this red tape and regulation will increase supply and drive prices down.
The key fact your missing is if you wave a magic wand and 10x San Francisco you wouldn't 10x all jobs. You would 10x the # of waiters, and garbage men but you wouldn't 10x the # of 500k/yr Google site reliability engineers. And it's the latter not the former that are driving up prices.
I mean, yes, by basic rules of supply and demand. But only in our world.
Most of America is very, very sparsely developed. There is just a plain huge amount of land with no one living in it, and another vast amount of land with very few people living in it. While on the other hand, there are very few Manhattans. There are more, but still very few, medium dense places somewhere in-between.
Some people want to live in more dense places, because they get the benefits of being near others packed more densely. While on the other hand, some people would prefer to live in less dense places, because they get the benefit of more land and space to themselves. And some prefer something in between.
Empirically, the demand for dense spaces is very high compared to the supply. And the demand slopes upwards with density, while supply slopes downwards. You can tell this is true because of the high prices of housing per sq ft in dense regions.
Whereas empirically, we are vastly oversupplied with distant rural zones where you can live far away from anyone else. You can tell this is true because of the low price of housing per sq ft in less dense regions.
This seems so natural to us that it's hard to imagine it being any other way. But that's just an artifact of having lots and lots and lots of land that is mostly empty.
Imagine living in a world where most of the land is super dense mega-packed city, far more than anyone can use, but where apartments are still plentiful as water. Where MegaManhattan goes on for miles and miles and miles and has so much housing that even though it's incredibly dense, there's still plenty of apartments vacant to move into.
And also imagine that relative to our world, there is a very small amount of empty land. Only a few places in this world where there is vast open space. Only 10,000 total homes exist, in this low density region.
Which is more expensive now: high density, or low density? Obviously in this other world the answer is low density. The demand for such low density housing would vastly outstrip the supply, whereas the plentiful apartments throughout MegaManhattan would ensure prices stay low.
There is nothing intrinsically expensive about high density regions, or low density regions. We just chronically underbuild high density by making it illegal to construct more, so prices stay high. If you make it impossible to build in a desirable low density region, you get sky high prices also (see: Aspen, CO, which is incredibly expensive vs its density).
I am not an economist, but I made an observation that got me thinking, and I came to a plausible explanation.
The observation: there are quite a few leftie environmentalists who insist that adding extra lanes to congested roads is pointless - the added supply will only induce more demand, so the road will stay just as congested.
Then they change the topic and demand that new, affordable housing is built in crowded cities. Hmmmmm.
So here's my simplified picture: smart, ambitious people move to big cities not because they want to, but because they have to, to find suitable jobs both for them and their equally highly-qualified spouse.
Big companies move to big cities because only there do they have a sufficient pool of smart, talented employees. Also, wealthy cities can offer good infrastucture and other incentives, and the wealthy bosses can shrug off the high costs of living and enjoy the cultural amenities.
Small companies move to big cities because that's where their clients are - the big companies and their well-paid employees.
Employees of small companies move there because it's still better to pay outrageously high rents than to be unemployed in a dead town.
All of this works the better, the more densely the city is packed (mostly in terms of "how many people can you reach with a 45-min ride in whatever means of transportation").
Nowhere in this argument did "it's fun to live in big, densely packed cities" enter the picture, except for the privileged few.
So as long as there are smallish cities and villages that the moloch monopoly city can suck dry, all that happens when you build another block of affordable housing in a big city is that the ratchet is cranked another notch toward "everyone lives in overcrowded, expensive urban hellscapes where only the richest few actually enjoy living".
How about we try to make smaller, less expensive cities more attractive to companies instead?
This sounds like a giant argument for a Land Value Tax. It is the only option which would increase building everywhere therefor decreasing home prices everywhere.
It should be very easy to convince you this is wrong:
So I don’t understand why Matt believes that building a few new apartments in some city - a very small move along that spectrum - would do anything other than make local prices go up
You are treating your model as though R2 is one. Your own data show its not. If there are cities both well above and well below trend line AND studies show small increases in supply controlled for other factors can reduce prices, then maybe take those results at face value. Or at least find data that show otherwise.
Theres no reason SD cant look at miamis dot and think, maybe we could get there with different policy decisions. It doesnt require a wierd divot in the curve.
If you want to be concerned that the larger scale relation cannot be improved, that is at least supported by your data. But movement along this curve is a long term change and multi factored and not strictly a function of how much upzoning happens in a given year
I think you’re conflating housing prices with amenities. Look at prices in Dallas vs Seattle. Seattle prices are higher than Dallas despite having fewer houses, because there is more demand, because there are more amenities such as restaurants, cultural activities, events, and high-paying jobs. This holds true even in small, expensive cities like Boulder, where there are popular amenities (beautiful outdoor activities, quaint downtown, etc), but not much housing. Large cities, by virtue of having more people, generally have more amenities. So there is a correlation, not causation.
To be clear, you may think that having more houses causes more people which cause more amenities, and it can be messy, but if that we’re purely the case we’d see no cities shrink while others grow, but we see that all the time.
Is induced demand is based more on the rank-order size of the city or the absolute size? I suspect the former is a big factor because there are dozens of large cities that don't even cross the mind of the sort of people who "would be happy to live in Seattle, or NYC, or the Bay". I count 56 metro areas above 1 million, and 15 metro areas above 4 million on this list: https://en.wikipedia.org/wiki/Metropolitan_statistical_area
There’s clearly a positive feedback loop causing urbanization. This is why cities exist in the first place and get larger and denser. (Contrast with farmers.)
The strength of the feedback loop varies by city and neighborhood. Some cities pull more people in and others don’t. We should expect higher-priced neighborhoods to have more of a draw for some reason.
Positive feedback loops can go on for quite a while but don’t necessarily go on forever. For example, demand for student housing from a university is limited depending on the growth rate of the university, and it might not be growing at all.
So I think this can’t be answered from first principles. You need to look at what’s driving demand locally. Will it continue?
I have long thought that dense California cities should be encouraging large employers to move, or at least expand more elsewhere. When an employer from a high-priced city moves to a city that needs more jobs, both places benefit.
"But Manhattan and London have the highest house prices in their respective countries, primarily because of their density and the opportunities density provides" implies a commutative relation that does not actually exist.
Yes, at the macro level, greater density -> more specialization -> more jobs/opportunities. But with transportation tech and cultural assumptions that commute time is the employee's problem, we have managed to severe the connection the other direction, such that we can create lots of opportunities/jobs/industry without also creating the resulting density.
I don't have the data handy to make plots, but there's a "natural" density of housing to match a given density of jobs within a metro area. A lot of our currently prosperous cities have made policy choices to prevent the housing density to grow to match the opportunity density. At that point, building more houses does *not* correspondingly increase the opportunities/make the city more desirable. It just enables the people in the existing jobs to be closer to them on average. If you built enough to actually start increasing opportunities again, then sure, prices will rise, but until then you aren't actually moving up that size/cost curve you started with.
I don't think you have justified excluding non-US metros, like Tokyo, or Auckland. Doesn't this lead to the natural conclusion that there is a sufficient level of housing to build, and that the problem is that the USA's many metros are structured to prevent housing? It seems like you're just arguing that US metros are bad at building housing, which is also what Matt Yglesias is arguing.
Your argument relies heavily on cherry-picked examples (New York City, San Francisco, North Dakota) and does not account for the broader housing market in the United States. This selective approach fails to consider the variety of factors that contribute to housing prices in different cities, and makes your argument less credible.
The assumption that density is the primary driver of high housing prices in cities like New York and San Francisco is simplistic and ignores the myriad of factors that contribute to their housing costs. Factors such as land scarcity, regulatory barriers, and high demand due to economic opportunities should not be overlooked.
The hypothetical scenario of Oakland becoming ten times denser is an extreme and unrealistic example. It fails to consider the gradual changes and adjustments that would occur in the housing market over time. Your argument does not account for the possibility that increased housing supply could eventually outpace demand, leading to stable or even lower housing prices.
Your claim that there is an inherent "coordination problem" between cities when it comes to upzoning and housing supply is unsubstantiated. You do not provide any evidence or examples to support this assertion, and it appears to be conjecture rather than grounded in economic theory or empirical evidence.
You fail to address the role of elasticities in the supply and demand relationship within housing markets. By not considering how elastic the demand for housing might be, your argument misses a key component in the analysis. If demand is relatively elastic, even small increases in housing supply could lead to lower housing prices.
Any way to analyze housing markets without considering the blatant and heavy handed manipulation of the real estate markets (either for gdp or property taxes) seems..... mildly optimistic about the price mechanism.
Mortgages are 30 year things and there's always some government program, theres going to be massive lag with unclear data at minimum.
Prices work best when the value of money is effectively linear, there are repeat costumers and products are interchangeable; which simply isn't the case for housing. You shouldn't assume that real estate has sane prices, and from deducing how a sane system would react to a change, conclude which effect.
I think you are using ill-fitting metrics and it confuses you.
Prices are high where the supply is less than the demand. Absolute values of supply and demand are not relevant. So the prices/density graph is not helpful. Most of the causality there is reversed and the second order effects depend on variables for which density is itself a proxy: amount of amenities and opportunities.
If the amount of amenities and opportunities in a place is increased, demand for houses will rise, thus housing prices also rise, and until possible new houses are build, increasing density. If you just increase density without creating new jobs, nighclubs, museums and other places of interests that will motivate more people to want to live there, the prices will fall.
Let me rephrase the argument: positive network externalities to building densely in cities are so large that building more housing actually causes existing units to be so much more desirable that its price goes up, despite the additional supply.
This might be true (it's certainly logically possible), but I suspect it's false on the current margin in most housing markets, even really weird ones like Oakland or Manhattan. This seems to be what the empirical evidence says, as far as I know.
Notably, if this were true the cost of overly restrictive zoning is significantly greater than the consensus view.
One example of where building a huge amount likely lowered price is Chicago's far northern lakefront just south of Evanston, where a huge number of ~30 story apartment buildings were built along Lake Michigan in, I'm guessing, the 1960s. By the 1990s, condos in them seemed pretty cheap.
In general, though, the U.S. doesn't have that many places where developers were allowed to run hog wild building high rises. In contrast, Sao Paulo, Brazil struck me as having an immense number of residential high rises by American standards.
Southern California, for example, has only a tiny number of high rises along the Pacific.
Someone on here, or another Substack comment section, said that the recent revealed American preferences in housing outside NY, and maybe SF, was a house with 3-4 beds and a yard or garden. Cities with those kinds of housing stock are expanding, the rest (excluding the two I mentioned) are declining or static. Looking at population flows in the last few years that seems to be true. California has lost 500k people to outward migration in two years. Surely you would expect house prices to fall given those statistics particularly if the trend continues - 2.5M in a decade would be a major decline.
Given that new reality Scott is probably wrong about Oakland and SF. Oakland seems to be an overflow city, the place to be if you can’t afford SF. Building new housing there now into a population decline would see prices fall. SF on the other hand may take up some slack.
What percentage of the earth's surface within, say 50 miles of the Dallas city hall could be developed for housing? I'm guessing somewhere around 95%, leaving out some lakes and the like. In contrast, what percentage within 50 miles of the San Francisco city hall could be developed? I don't know, maybe 50%? Maybe less. A huge fraction is saltwater and quite a bit more is rugged mountains.
Not surprisingly, land costs more near San Francisco than near Dallas.
1. density is desirable. This means that people mostly want to live and get higher amenities from a higher density city. I think this makes a lot of assumption on development that comes together with density - for example on public transportation, etc. But let's assume we get them with density.
2. There is a fixed amount of people willing to move to a metro in the US.
If you build more houses -> desirability increases -> each person in the city gets more from the city -> it attracts more people -> prices increase in the city, but reduce in the rest of the US metros.
Even if this is true, each person in the city gets more, so this is reasonable. Also, if this is done in all the cities (NY + Austin + SF + Boston + Auckland + Atlanta, etc) this will reduce the prices in all the US.
I spill 3 facts I know about urban economics and interesting top notch academic papers. I hope it helps.
(1) All else being the same, housing prices decrease when you build more (assuming same number of people, same wages, ...)
(2) Urban development increases productivity, wages and GDP/capita increases. (economies of agglomeration) As a consequence, wages increase and there are more jobs. More people move into cities. Commute times determine the limits of growths.
(3) Housing market equilibrium, and city growth is determined by multiple factors. Eventually negative externalizes will limit the growth. In the US housing markets, low density and bad public transport create negative externalities earlier than elsewhere. US economy would be much bigger if Urban planning and zoning would be better.
Two top notch relatively recent academic papers on the subject:
Hsieh, Chang-Tai, and Enrico Moretti. 2019. "Housing Constraints and Spatial Misallocation." American Economic Journal: Macroeconomics, 11 (2): 1-39. DOI: 10.1257/mac.20170388
Richard Hornbeck, Enrico Moretti; Estimating Who Benefits from Productivity Growth: Local and Distant Effects of City Productivity Growth on Wages, Rents, and Inequality. The Review of Economics and Statistics 2022; doi: https://doi.org/10.1162/rest_a_01208
I don't think the primary driver of demand is "wanting to live in a big city". I think the primary driver of demand is new york city wages. So if NYC built more housing the prices would drop as demand is met for all those who can achieve wages commensurate to the housing prices. Similarly if Oakland built 5 million houses they wouldn't see NYC style demand since there isn't NYC style wages
Maybe the relation between new housing construction and rent price go like this:
1) new housing -> rent price goes down (classic offer and demand)
2) more house -> new opportunity for services and shops, i.e. increase the productivity per m² of the place
3) increased productivity per m² -> more demand for housing ->rent price goes up (classic rent price depending on the place average salary)
So the apparent paradox is simply a 2 different effects with a time delay. New construction drive rents down in the short term, then up after some delay (if the city is succesfull, if not, it's a housing bubble crash)
Scott, I think you are missing the elephant in the room, which is jobs. If density is highly correlated (like the R^2 of 0.77 in your graph), Palo Alto should be cheaper than most US cities. The reason why it is not is because it is in close proximity to high quality, high paying jobs. On the other side, the reason why Detroit is doing bad for such a large city is due to the lack of jobs.
The causality is clearly: Jobs -> housing demand -> building. Yes, there are jobs that are created due to agglomeration effects, but for the most part, this agglomeration takes a long time. (Even Silicon Valley is built on top of 70 years of investments in tech). On the other hand, when oil is found in North Dakota, prices for small town housing rocket up.
Your thought experiment of Oakland was really weird, as you picked just a single city in a 10 million people metro area. There are enough jobs in the Bay Area that are already established for the area to absorb 4.5 million new residents. Applying your thought process to other cities shows the brokenness of the logic. If you magically turned Portland or Detroit into a 6.5 million person city overnight, there is no doubt that the price of housing would crater in Portland's case, and Detroit will definitely not become a new Manhattan.
If you accept that jobs create housing demand, then the solution is to keep increases in housing supply above increases in jobs. There are many ways of achieving this, some good, some bad. One would be to disallow cities from creating new office and industrial space without a corresponding increase in built housing.
If this is correct, then the intuition that Silicon Valley folks are increasing rent for everyone in SF is clear and logical, as they represent new jobs without the corresponding increase in housing.
> But it can’t do this forever - at some point, it will exhaust the pool of Americans who want to move to big cities (you’ll know this has happened when housing prices are no higher in big cities than anywhere else).
This is obviously wrong; most things are more expensive in cities than they are outside of cities for the very simple reason that people in cities have more money than other people do. It doesn't mean the urbanites want those things more; in many cases they want them less.
But even worse, you're treating a complex function of several variables, the price of housing... somewhere... as an indicator for a single variable, the size of "the pool of Americans who want to move to big cities". In the first place, that quantity is not a coherent concept: the number of people who want to move to cities is going to be affected by local conditions in the cities. If housing there gets cheap enough, more people will want to move than previously did. And in the second place, the general conceptual approach of judging the value of one variable by watching what happens to a complex composite figure that includes that variable is complete nonsense.
You are confusing density of houses with density of people.
If Oakland built 50,000 houses, that wouldn't make it attractive since people don't want to live in a town with 50,000 empty houses.
If Oakland built 50,000 houses and managed to fill 49,000 of them than that would make the remaining 1,000 more attractive since now Oakland is a bigger town with more amenities and culture. Although it takes time for amenities and culture to happen but yes, I would guess that if they built 50,000 new houses, managed to fill up 49,000 of them and promoted culture or waited for culture to happen naturally then the remaining 1,000 would be more attractive.
In terms of an "economic perpetual motion machine" this is not a big deal. I mean, any company that generates revenue and uses that to reinvest in the business is kind of like an "economic perpetual motion machine"
You could also do something like build 50,000 apartment units with great amenities, rent them out super cheaply for 5 years to make sure they get filled up and then once people moved in and culture and society is created jack up the prices. Don't know that anyone has ever tried this though since it sounds really risky. Building 50,000 apartment units is really expensive and you don't know if it will work. But the fact that "maybe there exist risky investment opportunities that if executed properly might yield massive returns over the course of a long period of time" doesn't seem to violate any kind of economic law does it?
Didn't Dutch originally settle in the area exactly because of the harbor? Wikipedia says
> The mouth of the Hudson River was selected as the ideal place for initial settlement as it had easy access to the ocean while also securing an ice-free lifeline to the beaver trading post near present-day Albany. Here, Indigenous hunters supplied them with pelts in exchange for European-made trade goods and wampum, which was soon being made by the Dutch on Long Island. In 1621, the Dutch West India Company was founded.
Surely geography helped New York later become a national capital and in any case a center of finance even after the establishment of DC.
You would obviously be wrong for any city in which demand is limited. If you found a city where nobody wanted to live, and put a lot of housing there, you would crash the price of houses in that city. You could use this idea to regulate housing prices in those kinds of cities - just build enough housing to lower prices to the level you desire, rather than crashing it. China has tried building cities from nothing, and many large buildings were later torn down having never been occupied. Even with as many people as China has, there are lots of places with existing housing available and nobody trying to live there.
The problem you seem to be identifying is that there are some cities where demand far outstrips supply, due to their nationwide pull. Lots of people want to live in NYC or SF who are unable to do so. This is coupled with the reality that developers are going to want to take advantage of that by building denser and more expensive housing, not cheaper and less dense housing. Because land is geographically isolated (you can't move the underlying land or the land next to it), there's only so much land available within a certain distance of whatever people want to be near. For SF it's likely tech jobs. Living a three hour drive away isn't useful if you want to work in tech on site. It's also not useful if you want to be a teacher or service employee in SF. Those could be great jobs, but you have to live close enough not to turn all your free time into a miserable commute.
I'm not sure if it's a counterexample or evidence for your theory, but Tokyo is notorious for being rather cheap for such a big, densely populated city. Their prices for houses in the 23 wards are about half of what you'd pay in the suburbs of Toronto (and Toronto in turn barely compares to Vancouver or SF).
On the one hand this could be explained away by the overall demand in Japan being capped. Tokyo is still the most expensive city in that country, and there's relatively few global migrants choosing between US cities and Japan ones.
On the other hand, Tokyo still has about 3x the population density of New York, so if they'd stuck to New York-style zoning, prices could have risen to absolutely insane levels (in the city itself and in whichever places people priced out chose instead). Conversely, New York moving to the density of Tokyo's metropolitan area (which is about the same size as New York itself), would support about 10,000,000 extra people. How big do you think the "any big city" pool of migrants is?
On the gripping hand, since the zoning rules are set nationally, it's not an example of "one city upzones drastically, but its neighbors don't". Maybe there would be significant extra demand if little-to-no housing supply could be built in Yokohama/Osaka/etc.
Find one of those little islands in Maine, then rebuild all of New York City on that isalnd in Maine. Are the new apartments in the new city worth as much as an apartment in NYC? Probably not because none of the people, businesses, institutions, etc in the original NYC that make it such a desirable place to live in are in Maine. There is a difference between population density and housing density. Population density makes housing prices go up. Additional housing make prices go down. Housing cost is the ratio of population density to housing density, ie the ratio of supply to demand like in any other market. So building additional housing shoud make prices go down in the short term, while long term you may have to deal with Jeavon's paradox whereby the increasing livability of the city makes it even more desirable to live in, causing an increase in prices, but that's not a given.
"No; I think the missing insight is that there’s some pool of geographically mobile Americans1 who are looking for new housing (or who might start looking if the right situation presented itself). These people have various combinations of preferences and requirements. One common pattern is to prefer any big city - they would be happy to live in Seattle, or NYC, or the Bay, if the opportunity came up."
Another common pattern is to want to live where you're from. In 1990 NYC had 7.3M people, compared to 8.8M now. How many of those 8.8M have lived here that entire time, or were born here during the interim? I was born and raised here, I like it here, and most of my friends and family are here. I'm definitely a big-city-simp, but Oakland would have a hard time getting me to move there by building 10X the housing for many of the same reasons that I don't currently move to Tokyo.
In the absence of zoning laws, housing costs will equal the cost of construction. If you get extremely dense like NYC and run out of land you'll need to build way up, which is very costly.
But most places in America can get 10x denser without having to build more than 3 stories high. If the area has housing prices higher than the cost of construction, eliminating zoning laws will lower housing prices even while inducing demand; you can just keep building to accommodate new arrivals!
As an analogy, we induced demand for cars by building lots of cars and roads. But cars didn't get more expensive, since we can scale up production to meet demand. Car prices will only increase if the cost of construction increases (albeit with short term fluctuations from sudden spike in demand).
Cities in the past were not this expensive for that reason. Chicago grew from 4,000 to 3.4M in less than a century, but managed to accommodate everyone.
I think more housing increases prices because it's limited. Looking at San Francisco, the city proper isn't even a million in population (the metro area is 7.7 million).
I have no idea how many people want to move to San Francisco, but let's say that it's 100,000 out of the entire world (because you'll get the best jobs in Silicon Valley and get rich).
If SF builds 90,000 new housing units, that means 90,000 more people can move there. So it's not going to bring down prices for those already living there, who might like to move out of a houseshare and get their own place, since the demand is continuing at the same level. And the price won't come down, because the housing supply isn't at the same level as housing demand - there are still more people who want to live in SF than there is housing available.
So increasing density won't bring down prices in that case. Unless you can say "In 2028 the projected population will be 2 million and by then, with old and new housing stock, we will have 3 million units available", then the price will come down. But so long as more people want to/need to move there, so that new housing is being snapped up as soon as it becomes available, then prices will remain high.
Same for the metro area - so you can't afford to pay rent to live in SF proper, then you move to the nearest city/town where you can pay rent. Soon everyone is taking that as their second choice, and demand is pushing up prices. Then people switch to their third choice, and so on, until you get to the places where the distance in travel time is so far, and there aren't commuter trains, that it's not in demand and so prices are lower.
I’m not entirely sure I disagree, but this also seems like an unnecessary addition that doesn’t explain much--particularly as it is contradicting by empirical studies.
Here’s what I see happening:
1) Land prices increase when an area is unusually productive or desirable.
2) This then incentivizes more housing units to be built to take advantage of that additional productivity/desirability.
3) Reactionaries (not necessarily a bad term here) oppose further construction, leading to continued increases in price.
There are two additional effects, one of which you posited
A) As density increases, the desirability of the city for urbanophiles increases.
B) As density increases, productivity increases because there are more talented people in a smaller radius and because you can tailor your local business to a smaller niche of people while having the same number of customers.
Your argument that ‘A’ is a main factor in driving prices up is fairly unconvincing to me. Plenty of people, if not most people, seem to prefer the suburbs to cities, at least aesthetically. On the other hand, ‘B’ seems fairly likely.
But what I want to specifically point out is that your argument rests on the claim that the cost increases from section 1 and 3 are smaller than the cost increases from section A (and B, if you include that).
Urban desirability and increased productivity (A and B) are positive feedback loops whereby increased housing leads to increased demand and therefore increased prices. However, ordinary supply and demand is a negative feedback loop whereby increased housing leads to increased supply and therefore decreased prices.
One important difference between these two models is that, if B is driving the increase, local wages should be expected to actually grow faster than local housing prices even if said housing prices increase, because the attracting of productive talent and productive companies (combined with Baumol’s cost-disease) will cause wage gains across the spectrum.
The only losers in this scenario are individuals wealthy enough that they would have preferred constant wages/salaries in exchange for continued suburbia.
a) relaxed housing restrictions increases supply which decreases prices and rents
b) increased density creates more valuable urban agglomeration effects
c) urban agglomeration value is capitalised into valuation of dwellings, increasing demand and increasing prices
d) c) is bigger than a), so the net impact is that prices and rents rise.
The problem is that part d) cannot be inferred from the data (that would be) behind the imaginary chart because each of the cities on it have restrictions on dwellings that we aren't controlling for (and would have a hard time trying to work out exactly how to control for in any real-world study).
It's also not quite the right framing to think of a reduction in zoning restrictions as in increase in the supply of housing. It is really an increase in the elasticity of the supply of housing - by reducing absolute legal restrictions as well as the implicit (though sometimes more restrictive) barriers from the approvals process, the supply of housing becomes more responsive to changes in prices/rents.
The simple conclusion in theory (for instance In the Rosen-Roback model of spatial equilibrium) is that, whether urban agglomeration (which is effectively a kind of productivity) winds up in prices and rents depends on the elasticity of housing supply. Where the housing supply is perfectly inelastic (i.e. no price increase can induce the supply additional housing) all productivity increases go to house prices. Where the housing supply is perfectly elastic (i.e. an infinite amount of housing can be supplied at the going price) all productivity increases go to wages and population growth.
The imaginary graph (from the article) could be explained by the following stylised set of facts: historically freer planning, but urban development commencing at different times (and at different rates for random "natural" reasons) followed by much uniformly strict restrictions on redevelopment everywhere. The initial level of density (and thus urban agglomeration) is different in different places, but with similar levels of restrictions on new development (and additional densification) housing supply is similarly inelastic everywhere and so prices today are largely determined by density levels. This stylised history is consistent with frequent observations that, much of the development in big cities would be illegal today (e.g. https://www.nytimes.com/interactive/2016/05/19/upshot/forty-percent-of-manhattans-buildings-could-not-be-built-today.html)
The reality is likely a messy version of this (because restrictions are different in different cities and have been implemented at different times). But the important lens is that the relationship between prices and productivity (and thus density) depends on a relatively inelastic supply of houses - exactly what is being alleviated by zoning reform and relaxation.
The other insight that I think is missing from Yglesias' (and your) analysis here is that growth is multifaceted. Density doesn't increase for no reason; it increases as a knock-on result of growth in jobs, and amenities, and clubs, and service organizations, and sports teams, and etc. These things are really, really heavily coupled.
When cities go through periods of "growth" meaning the broad definition above, housing demand tends to grow faster than housing supply, which leads to things like above-market bidding wars and spiraling rents. *That* leads to density. Building stuff takes time; the density follows many months to multiple years *after* the increase in demand.
So, I disagree with "density increases local prices" per se; *growth* increases local prices, which prompts denser development. Where my YIMBYism comes in here is not that you can make a desirable city cheap with lots of new housing - sorry, not gonna happen - but rather that policies which restrict density should be avoided precisely because they guarantee that the price shocks that come with growth will be sharper and take longer to resolve.
Part of why this is all so important is that when price shocks do settle, they don't resolve into big price drops - price increases in housing are very, very sticky. Homeowners *hate* selling at a loss, and will avoid it like the plague. Landlords won't cut rents until vacancy rates become extreme. Policy has to be designed to keep the price jumps under control, because once they occur there's no path down that isn't super painful.
It sounds like this is just a matter of accounting for covariates. Housing density isn't the only thing that makes an urban environment valuable to live in. There need to be jobs, transportation options, food availability, basic services like plumbing and trash collection, a lack of crime, weatherproofing, public spaces, amenities, and so on, all of which we might lump into the attractiveness of a location.
If you look along the vector between the random plain in North Dakota and NYC, both the attractiveness and the housing supply increase, and the difference in price indicates that the attractiveness increases more than the housing supply does. But when it comes to discussing changes to housing policy in isolation, you're operating on a different vector that keeps the attractiveness more or less constant.
A simple graph might help, which I can describe in words: Put attractiveness on the vertical axis, and housing supply on the horizontal. Price is highest in the upper left, lowest in the lower right. NYC is above and to the right of ND, but it's more above than it is to the right, so along that diagonal, price increases as you go up and to the right. Housing policy debate is about approximately horizontal changes, and on that axis, price decreases as you go from left to right, i.e. increase the housing supply.
Scott, I'm tempted to say that you're not even wrong, but in an interesting way.
Your theory is that increasing housing density in places where people want to live increases prices, but that it does so in increasing the value of living in the city as a whole. If you're right, and I think maybe you are, then it doesn't make sense to say "We shouldn't build housing because although it will increase supply, allowing more people to live here and lowering prices, it will also make almost everyone in the city better off, therefore increasing demand to live in the city and increasing prices!" If you're right, we should build more housing both because it allows more people to live in the city and because it improves the lives of nearly everyone in the city - there's more housing AND each unit of housing is more valuable to the people living in it than it was before.
Quibble one: I don't see anything in your essay that suggests that the increased value effect would raise overall housing prices more than the increased supply effect would lower them. Do you think this is inevitable or even likely, and if so why?
Quibble two: Of course, you might still oppose YIMBYism because you were concerned that increased density would change the character of a city in undesirable ways, or you might be concerned that particular groups would lose out on the increase in value provided by more housing.
Adding housing only increases density if the land area is fixed. If new land is annexed for the new housing, available housing can increase with no change to density.
And the units in the first graph, density per unit area, make no sense to me.
There is a difference between house prices (which include the land under them and so dramatically increase as density increases) and housing rental prices (apartments/condos can be stacked to make use of the pricey land). That being said building up is harder than building on a flat plane, so I would expect the rental prices to increase relative to a flat plane (if somehow building a 20 story building was cheaper we’d see them built in the middle of nowhere more). So density does inherently raise rental prices somewhat because it requires building up, but artificial restrictions on building density will increase prices even more than they would be without them. I think it’s also right that density can increase demand for a particular city as some people want to live in with the extra amenities of density (and tolerate/enjoy the density itself) vs people who want to live in open country or more suburban areas. Japan allows more density and construction everywhere (plus falling population) so their housing is just a whole lot cheaper overall, but still should be more in Tokyo than in the country due to the height. As with many things in Japan, I suspect they are probably more at an overall equilibrium with regard to people distributing throughout the country and having the interior/exterior space, and density/amenities for their income vs the US which is much more dynamic, has a great many more building restrictions and inefficiencies, and has eye popping income increases from the tech and oil industries.
We also do things like requiring double stairs and have worse fire rescue capabilities, which increase the cost of building up vs building flat in the US relative to many other places. Property taxes/Georgism can significantly affect this as well, not sure how that differs between the US and Japan. There are also significant cultural differences in how income is used and wealth is saved, with the US using housing as the way to build a typical household’s wealth.
Didn't Robert Lucas point this out in simple terms like a hundred years ago when he noted that cities are valuable because of the social connections and the more people/housing units, adjusted for proximity, the more social connections, and therefore the more valuable the housing unit or the restaurant table.
I don't think your graph of price vs. density is as ridiculous as it looks at first glance.
You can very well make a model where each single added house decreases local prices by a small amount while having an overall positive correlation between houses and price in aggregate.
You just need some event that is more likely to happen with many houses that increases the price by a disproportionate amount, like a school, supermarket or public transport being built or a high profile business moving to the area. Or, simplyfying the model, an event that happens once a certain threshold of houses have been met. The resulting graph would go down with each single house but would also jump up at certain points, thus increasing the price in the long run.
Of course, this is a very YIMBY-favoring interpretation. Ignoring other factors, it would favor YIMBYism even on a geographically local scale from a community perspective: Prices (mostly) go down because there is more supply of equal quality and prices (mostly) go up because there's more quality of equal supply.
For inhabitants of big cities, who can actually afford it now, it is undesirable to decrease housing costs in big cities, because that leads to an increase of poorer people in their vicinity.
Considering cities, I think its not a function of big cities attract people. No, its jobs and amenities.
100K city no airport. no art museum. few high paying jobs
500k city - has airport, has art museum, has ethnic food, has a high tech business
NYC level - has stock exchange and banking industry able to support top class amenity, also other jobs, that attract other amenities. OOOH, HAKKA FOOD. OOH, AN OPERA AND A SPORTS TEAM.
You're not adding houses to a generic city with no special amenities. As the city gets bigger based on industry it also gets more amenities and becomes more attractive. OOH BALLET, OOOOH, ARTISAN PICKLES.
I haven't even considered cachet or prestige!!!!
But basically this will vault certain cities very high up for high high demand.
Adding a few more houses might not be able to satiate the pent up demand that is so high, these urban locations have people commuting for an hour on a train to work there.
I think you're onto something. But you should check out this paper by Ed Glaeser and Joseph Gyourko: https://pubs.aeaweb.org/doi/pdfplus/10.1257/jep.32.1.3. From Figure 3, housing prices in most cities are pretty close to the costs of construction in those cities. Deviations from this rule are cities where people want to move but there are natural or artificial barriers to housing construction.
Your story is based on the idea that density carries attractive amenities. Anecdotally, I think the amenity value of city density has been enhanced by recent technology. For example, technology has greatly eliminated the need to wait in stupid lines (at the bank, at the post office, etc.). And when you do have to wait in lines, it's not nearly so bad because your phone allows you to still be productive and/or entertained. Technology has also made it easier to navigate a dense city and to find all those niche little shops or clubs or farmers markets or whatever that make a dense city so cool.
Strictly speaking - density is irrelevant. What matters is demand vs. supply.
The housing building advocates use this incorrectly: they presume perfectly matching demand vs. supply - thus increasing supply reduces demand effects. Wrong.
First of all, no city is standalone. San Francisco has the entire Bay Area as a suburb; New York has its other, non-Manhattan boroughs as well as New Jersey, Washington DC has Maryland and Virginia, etc etc. These suburbs/adjacent areas constitute both latent and unmet demand - while not everyone in those suburbs wants to live in the city, the reverse is not true at all: few people in the city want to live in the suburbs. The net demand effect of these suburbs is clearly skewed towards the city. In this respect: Yglesias is completely correct in that building 1 or a dozen apartment complexes will do nothing to materially affect the supply/demand curve.
The second problem is that there are material effects from social mores and government policy. Both San Francisco and New York City have property tax caps, for example. These caps serve to reduce turnover because anyone who has owned a property in those areas has a material tax benefit that increases over time - they only sell if they have to and/or are moving out forever. There are numerous properties in Pacific Heights in SF which are $4M to $7M in value but are paying under $3000 a year in property tax (use Redfin to look around); the identical properties, if purchased today, would be paying at least $40K to $80K a year in property taxes. New York city has only had the tax cap since 2011 but the effects are similar, if smaller. Social mores - Tokyo is a prime example. People simply live in smaller places. San Francisco and New York city also have rent control - another market distorting mechanism on top of (and directly related to) the property tax increase limits.
Lastly, cities are where rich people live. Yes, some live on huge ranches in Montana or in a "regular" house in Omaha, but the majority of rich people live in cities and/or have one or more properties in a city. Rich people = more expensive property.
I find Scott's argument compelling, but to me many of the comments seem to be arguing about tangential points. So I wanted to try to reiterate what I thought was the core of the argument, for what it's worth. Less eloquently than Scott, alas. And then I'll try to offer a slight extension.
In this simplified thought experiment, imagine all housing units are identical, so we abstract away from the effect of changing housing quality. And imagine employers' location choice and employees' location choice are governed by the same factors, so we can think of a representative person's preference.
The core assertion: suppose you build 1,000 new housing units in San Francisco. This will probably cause _more than_ 1,000 people to move to San Francisco (specifically, people who would not have moved to SF in the absence of this new construction), and so housing prices will probably go up.
I know it seems weird, but I actually find the assertion pretty compelling, because the value of the network is proportional to the number of nodes in the network. If you build 1,000 new housing units, and fill them with 1,000 new entrants, doing whatever cool things those new entrants are doing, does that make the city more or less attractive to the marginal new 1,001st person deciding between SF or LA? "More" does seem like it has to be the right answer, most of the time. So if someone was indifferent between SF and LA given (old population, old prices), then they must prefer SF with (new population, old prices), since it now has a superior network. And hence equilibrium demands that prices go up until indifference is again achieved.
If I had to think about what might be missing from this picture, the only thing that comes to mind is that the intertemporal behavior of the system might be important. Rather than building 1000 new units, consider what happens when you build the _first_ unit. The 1st new entrant, in the old steady state, did not come to SF. At (old pop., old price), they chose LA. And while you now have a new unit, you still have the old population. So to change the first person's mind, you have to offer a lower price.
So this suggests we should expect, if a lump of new housing appears, it should have a short term negative effect on prices, and a medium-to-long term positive effect on prices. Which also makes sense because new entrants to a city don't immediately start contributing to its network effects; they have to be integrated into the network first (i.e., the famous celebrity chef entrant has to open their restaurant, the wealthy angel investor has to become known to the entrepreneurial ecosystem, the artist has to open their gallery, and so on.)
One last generalization. If the short-term and medium-term effects of a new housing unit on house prices have opposite signs, then in a continuous system, the relationship between prices and building rates will probably be non-monotonic. Think of a differential equation where changes in prices are a function of (new entrants - new units), and where new entrants' rate of entry is a function of current prices and _lagged_ population (where the lag is proportional to the amount of time it takes for the average person debating between SF and LA to update their decision model to include the network value of recent new entrants.)
So you can have one steady state with no new units, no new entrants, and no price changes. Or another steady state with very slow building of new units and increasing prices (because the rate at which new entrants are integrated into the network exceeds the rate at which new units are built). Or a third steady state with fast building, where units are built faster than new entrants are integrated, and we observe decreasing prices.
This could conceivably make sense of both SF and Tokyo.
Plot price change since 2000, say, against population growth over that same period. It will be immediately clear that there are no shortage of slow-growing cities. Control for average wages and it will be immediate.
I live in Whitehorse, Yukon. The average home here costs 50% more than the average home in Houston Texas. Yukon is about as sparse as you can be. It's got more land than Germany and the whole population wouldn't sell out the stadium where the little league world series is played. Density we ain't got. NIMBYs and city planners we got in spades.
Scale matters a lot here, so I'm confused why you want to disqualify the Tokyo example (or Seoul for that matter). Tokyo and Seoul make up a huge fraction of the populations of their respective countries (20-25%). It's true that _locally_ there are agglomeration effects so prices may move up, but if some place is YIMBY enough prices will move down. As far as I'm aware, _every_ locality that builds at the scale that the full-on YIMBYs want them to build as reduces prices.
You may disagree with turning the bay area into Tokyo. But I don't see how if the bay area built enough housing for 66 million people (20% of America) that prices wouldn't go down.
You excluded Tokyo from your dataset. Tokyo has much higher density than SF and much lower price per sqft.
I think there's a significant chance that this is a weird effect of modern zoning (or maybe service economies?) cities used to be where poor people lived! *Downtown* used to be full of poor people! There are major cases of high-density dirt cheap living (like tenements). It seems at the very least plausible that
NYC/SF are expensive because there are MANY good jobs there and people WANT to live there. Not because of the density of housing. You could build 500,000 homes in the middle of your empty field in North Dakota, and it wouldn't do much for the demand there. You aren't going to create Manhattan by magicking 3.5 million housing units of similar quality into the Red Lake Indian reservation in Northern Minnesota.
But overall you are right, that in the real world local density increases can/often lead to local gentrification/rising prices, and the benefits tend to be for the wider area. But the actual impacts depend a lot on the particulars and the services/jobs available.
Take it to an extreme: I have built an apartment in a pocket dimension (TARDIS-style) in New York. It has effectively infinite capacity. What will this do to New York real estate pricing? Do we expect that the apartments next door to mine will become more expensive as a result?
This is not actually Matt's view per se.
The important thing is to uncouple housing and land. Density increases the value of the land. But it also allows more units to occupy an existing parcel of land, which decreases the per-unit cost of housing.
So here is what you expect to see happen when new units are added at the margin:
1) Cost of all nearby buildings go up (this is actually the land value increasing, but we usually bundle land and building)
2) Value of existing single-family homes increases for the same reason as (1)
3) Cost to move into the neighborhood goes down because there are now more places to live.
It doesn't always happen precisely this way, but this is broadly what you'd expect.
You need to disaggregate housing and land.
In a perfectly competitive environment, price of housing (not land, the literal house) would be driven down near to what it cost to build. Denser cities usually have more productive people, so labor costs are higher.
Land is also fixed, so that does scale with amenities. Land close to amenities is valued higher, all else equal.
In this theory, how would something like California upzoning everywhere work? Would the local effect drive prices up? (California is one of 50 states) Or would the global effect drive prices down? (All cities doing it together)
Same question if the US upzones. Same question if a city upzones. This local vs global dichotomy seems poorly thought out.
Doesn’t it depend on the link between population and housing? Manhattan isn’t Manhattan because of the number of apartments but because of the number of buildings (and all the services and amenities that follow them).
I doubt the link between the two is as strong as you’d think; there may be a few people who don’t move to a city (think metro area, not urban core) due to house prices, but I suspect they mostly just live in worse (smaller, less convenient, further away, more crowded etc) housing. If you build more housing in Manhattan, you might suck people in from Brooklyn or Jersey City, but I’d bet on it lowering prices in Manhattan’s internal housing market, and Greater New York’s, because you won’t suck more people into the metro area.
In other words, I don’t think housing density causes population density; I think population density snowballs itself, and tends to cause housing density as a byproduct.
Imagine we built 10x houses _everywhere_. What would it do to the land price? I imagine that the total price of all land would remain more of less the same, but the price in the most popular places (Manhattan) would rise because it would be possible for more people to live there. It seems like it implies that the prices in the less popular locations should drop.
Of course I may be wrong about the grand total remaining the same, after all building more housing will create value.
Everyone is bringing up Tokyo, which was my first thought, so I'll instead mention that London is a place that's experienced quite a lot of growth and while it's pretty dense, it's also not very easy to build new stuff there which is why rich people build big underground complexes under their existing houses. For that reason it doesn't seem like a good example of YIMBY not working.
The key empirical facts that go against your story is that physical mobility has declined in the past decade while housing prices have shot up. The tiny share of hyper mobile people you're thinking of aren't nearly enough to move the needle for cities. In general, people move to a city they have no connection to for jobs, not housing.
I think one intuition you're missing here is that allowing for easier construction of housing at the policy level in general shifts housing costs closer to their actual construction cost. And, higher density housing is (to a point) cheaper to construct per unit than sf housing.
I asked an academic about this because I was interested in what the effect taking zoning controls off in areas marked R-60 (single family dwelling) as the local developers and their supporters kept saying it would help develop affordable housing (yet at the same time it seemed like everything new getting built was more expensive than the stuff it was replacing). Locally it does drive up prices. According to my colleague, in the wider DC-metro area it should not because areas in which there were 'old' housing stock have to reduce prices to remain competitive with the 'new' build areas. So Chevy Chase isn't going to get affordable, but mid-South East might see a price drop. At least that's the theory. There's a good example of this in Minneapolis (as they took zoning controls off in 2018). See
https://www.bloomberg.com/opinion/articles/2022-08-20/what-happened-when-minneapolis-ended-single-family-zoning?cmdf=minalpois+zoning+controls+result#xj4y7vzkg
and the effect of Covid with WFH now means they are going back to the drawing board again.
https://www.minnpost.com/cityscape/2023/03/minneapolis-2040-zoning-remains-stuck-in-the-1960s/?cmdf=minalpois+zoning+controls+result
I think this is missing an important piece of the puzzle. There's a big potential difference between the number of housing units in a city and the number of people living there. If Oakland had 4.5 million housing units suddenly appear that wouldn't summon an additional 4.5 million people into existence into the city.
The thing that makes New York a more desireable place to live than some island off the coast of Maine isn't the amount of housing there, its the number of people. And not just the people, the pattern of economic activity too. If you created those 4.5 million new houses in Oakland and just teleported people in randomly from across the country they'd all want to move out immediately because they wouldn't have any job to do in this hypothetical Oakland and the things that make a city a pleasant place to live like theaters and restaurants wouldn't exist either without a big shaking out.
If you just created 4.5 million new homes what would happen would be that maybe a few tens of thousands would move in immediately or lets be generous and say 500,000. Then the people who owned the other 4 million would have to slash rents to try to not stay empty and you would see prices plummet enough to start attracting people from all around the Bay Area driving down prices in that metro. But most people have a job or friends of family that prevents them from moving across the country just to take up residence in some apartment that happens to open up.
I think you're making a very common mistake here of confusing supply/demand with *quantity* supplied or quantity demanded. (This is very common! we teach students about this in micro 101 because it's so easy to make!)
What you're seeing is that the quantity supplied is correlated with housing prices (true!). But this is very different from establishing that the supply curve--i.e. the amount of housing that would be produced at any given price, and what moves up/down when we regulate/deregulate supply--is positively correlated with price. Figuring out what supply curves look like is a lot less intuitive and requires some high-grade econometrics, which is why economists had to set up a whole commission just to study this particular problem (the Cowles Commission).
In terms of resources for understanding how these concepts are different, a micro 101 textbook will cover this distinction. For the econometrics side of this, I've heard good things about Scott Cunningham's *Causal Inference Mixtape*, although I haven't personally used it.
I'm not really grokking your argument here.
There's a number of people who would move to New York if the rent was lower, but it's not an inexhaustible number of people. Keeping housing limited means that the price of housing stays high. The only way to get it down is if supply rises to meet demand. We know that there is an artificial constraint of creating more supply in the housing market in these cities, so it's no suprise that supply is not meeting demand. Every apartment you build is a step closer to satisfying the demand that already exists.
I think that's the issue. The demand is there already, building housing doesn't create demand, it just allows the demand that already exists to be satisfied.
I don't even know what to say further than that. I'm just don't get the induced demand argument. You can't lower the price of something by restricting supply, that just doesn't make sense. London and New York have expensive housing because they restrict supply, not because they encourage it.
I think what one also needs to consider is that $1 is in a sense worth less in NYC than in North Dakota – everything is more expensive in NYC but people also have higher incomes. (I think similarly, 1€ is worth less in Germany than in Latvia. It must be like that in order to balance supply and demand of things from Germany and Latvia, or something like that.)
So, I wonder what that housing price graph would look like if you corrected for that effect.
Price of a house is a function of 2 things: price of land and price of constructing a drywall box with a toilet. Upzoning Oakland definitely wouldn't drive the price of land down, but it would split the land price more ways. So instead of 1 person paying for a 5000 sqft lot, you could have 10 people paying for that same 5000 sqft lot. Even if the price doubles, the price per person goes way down. The cost of construction is higher for building tall skyscrapers vs building single story drywall boxes, but the difference is on the order of 1.5-2x, so as long as the cost of land makes up more than half the cost of your house, then you have margin to decrease that price by building taller. There's probably in theory an optimal density given the supply/demand and construction costs for each parcel and in Oakland, the optimal density is clearly much much larger and best determined by the free market, not city bureaucrats.
It seems to me that the "density increases prices" is not a direct effect, but part of a feedback loop. People are not attracted to more people per se. But to the things those people create (jobs, parties, theaters etc. ). If you just build a lot of houses you won't immediately get these effects, so you can't turn a small village into New York overnight. If you build new houses you get a temporary drop in prices. Untill the new people that live in these houses have allowed for more development to occur (they build a new buisiness or they are customers for the new theater). After that prices increase again. If you then build new houses the prices will lower again. So constant development of houses result in lower prices.
The YIMBY interpretation of the positive association between price and density is that virtually every city in the country adopted a strict zoning regime sometime in the mid 20th century, and the degree to which these laws caused a supply squeeze is mediated by density.
To falsify that we’d need a dense city to drastically liberalize development such that supply spikes upward and see what direction prices move. There aren’t really enough examples of this yet. Outside the U.S. there’s Auckland, which seems to support the YIMBY thesis. But not much that qualifies domestically.
Possible typo: “ But if becoming just as big as Manhattan or London would make Oakland more expensive, shouldn’t we assume that a little step in that direction would make it a little bit less expensive?” Shouldn’t that “less” be another “more”?
You lost me at "So I don’t understand why Matt believes that building a few new apartments in some city - a very small move along that spectrum - would do anything other than make local prices go up." How would they make prices go up? The would just accommodate the latent demand (unless those newcomers have somehow a stronger preference for city housing and willing to pay in excess of existing ones). Of course, one may observe an average increase because the new construction is probably of higher quality/more luxurious, but presumably you want to look at quality-adjusted prices.
I think your mixing up the agglomeration effects of density, which is what induces the demand, and the housing supply. You can't just build a city and expect people to move in, China has tried that. But if you have the agglomeration effects of density and shortage of housing due to artificial constraints, which we have all across the US, then you get dense areas with high housing costs.
Tokyo and NYC both attract tons of new residents But Tokyo's housing rents have been stable, while NYC rents keep rising.
Why? Tokyo has permissive housing construction laws. NYC makes building new housing almost illegal.
Yes, dense cities are attractive, and that makes them get more dense over time. But it only makes them more expensive if you forbid new housing to keep up with the new residents.
Wouldn’t there be a lag effect?
Should increasing the supply of housing not induce demand until that supply had been soaked up and the new people have started doing exciting things that induce demand?
So as long as the increased supply keeps ahead of the demand then housing prices shouldn’t go up, but go down by some proportion
There's two things going on here: confusing shifts in demand with movement along the demand curve, and getting causation backwards.
You're assuming density causes prosperity, rather than prosperity causing density. There are ways the former can happen, but the bigger thing is that, for a wide range of historical reasons, you can make a lot of money in NYC and SF, so lots of people want to live there, so they get very dense. This is the prosperity shifting demand right, so at any given price, more people want to live there; this drives prices up, and they go higher the more fixed supply is.
If you built a bunch of housing in Oakland, lots of people would move there because it's cheaper, which is movement along the demand curve; it's still the same number of people who want to live there at any price. Now, it's possible that the increased number of people living there makes the city more prosperous (this is the phenomenon of induced demand), which would shift demand right, but there are way more differences between NYC/SF and Oakland than just the density, so I don't think it would shift demand enough to offset this. In particular, if it's just a small increase in small, it's also a small increase in density, so there's almost no shift in demand (but there is movement along the curve).
"Lars Doucet, please report to the comments section. I repeat, Lars Doucet, please report to the comments section."
Tokyo! But I’m like the 10 person to bring it up…
Strikes me that the missing factor in this analysis is jobs/industry.
Consider two scenarios: in one (Scenario A), a city permits a bunch of new apartments, but no new offices. In another, (Scenario B) a city permits a bunch of offices but no new apartments.
Obviously the simple existence of an office doesn't guarantee a tenant...but it doesn't hurt (more available offices means it's easier to start a new business). Lots of offices means lots of office workers, means lots of people with professional salaries hunting apartments...I don't think it's crazy to argue that apartment rent is going to go down in Scenario A, and up in Scenario B.
And of course none of these cities appeared from nowhere. You start with a favorable port (usually). Some businesses supporting the operation of the port; then some industry taking advantage of the port; now services for all the people working at that stuff; now corporate headquarters for the growing businesses here. And so on and upward as the network value of the business and industrial cluster goes, achieving true takeoff when you get a New York or London style global financial hub, or a San Fransisco tech hub.
There's clearly a back-and-forth dependence with housing there; the business cluster can't grow if it can't find workers. But I'd argue the housing cost calculation isn't more housing -> bigger city -> more desirable -> higher prices. It's more like industry cluster -> more industry -> more desirable -> higher prices. And critically, that suggests that at any point in that industry network effect takeoff you can add new housing at a rate that will bring down the price of an apartment, because the new apartments aren't the critical factor in driving the business cluster value spiral.
Or to put it another way: NYC isn't the most expensive place because they built the densest housing, NYC is the most expensive place because of the network effect of being the business and financial hub of the richest economy in the history of the world, coupled with regulations on housing artificially restricting access to living near that business and financial hub.
Isn't the Tokyo example just what happens when you exhaust the pool of Japanese people who want to live in big cities? (Or maybe not exhaust -- my guess is that Tokyo is more expensive than random Japanese small towns -- but the gap is not as large as in the US because the fraction of the population wishing to move is smaller). The reason for this is that Japan's population hit a peak at some point in the late 21st century and has since been shrinking; I think if every new Japanese generation was larger than the one before it prices in Tokyo might still be soaring.
"Change my mind about housing, but don't mention Tokyo" is like saying "Change my mind about gun possession, but don't mention Switzerland."
You can't test the effect of allowing new housing unless you're willing to look at cities that do, in fact, allow it.
Most cities around the world block new housing. It's a popular policy!
And that means looking at cities subject to state-level permissive housing policies, like Tokyo.
You're assuming that cities want to decrease prices. Palo Alto homeowners only care about 1 thing in life: the price of their home. If they could drive up the price of their home by building more, they absolutely would do that. In reality, they know that building more homes would mean more competition for them and also nicer, newer competition. Who would buy a 1200 sqft home built in 1950 when you can get a nice new 2000 sqft townhouse for less?
I’m no economist but it seems there are a few things going on here:
1. Constraining supply when there’s high demand increases prices: this is the main YIMBY argument and I think it makes sense.
2. Currently, dense places are also expensive. Sometimes this is used to argue density doesn’t decrease prices, but surely part of it is about a third variable, namely demand. Again I think this is the standard YIMBY retort to the observed positive correlation between density and prices: nyc isn’t expensive because it’s dense--density and cost are both responses to demand.
3. Demand is driven by a bunch of things, likely including a high density of people (extent of the market, etc). I think this is largely the point you’re making here, right?
I agree that the relationship is probably complex because demand isn’t constant and (possibly) is driven in part by supply. But I think even that could be an oversimplification. People don’t just move to dense cities because they have lots of housing units, they move in the search for opportunity (economic and otherwise). So building more supply per se doesn’t necessarily increase demand: in large part it meets existing latent demand, as per 1 and 2. But to the extent 3 is true, then it’s probably also true that: increasing supply satisfies that demand, increasing number of people, which could in turn increase demand as per increasing extent of the market. Which I think is the feedback loop you’re identifying.
I just don’t think it’s a clean relationship or clear feedback loop though: increasing housing units doesn’t predictably increase demand, presumably--otherwise non-superstar cities like Detroit could just build more housing and would see increased demand and economic growth; but that’s not the relationship we see (afaik). This implies that a large part of demand is independent from (not caused by) housing stock, which is also what a YIMBY would argue, I think. Which means that to the extent supply is limited and demand is high, building more supply is important for satisfying that demand.
But I don’t know. I’m just speculating and this admittedly feels difficult to tease out empirically. It’d be a start to draw some DAGs, though.
I thought this was all caused by network effects.
Maybe I'm crazy but I thought I remembered an old Krugman paper on the positive network effects of large urban areas, written to explain why the trend of suburbanization was reversing in the 80s or 90s.
Like, there's nice things about SF and the Bay but most people want to be there for the tech scene. If you're a startup, it's worth paying $80k/year for the founder to live in SF because all the VCs and talent is there. Let's say that benefit is worth $100k/year over living in North Dakota. Alright, so if you build 20% more housing and 20% more startups and VCs move in, say decamping from Denver, then regardless of what the rent is, it's now worth $120k/year for the founder to live there because there's more VCs and more talent and more support there.
Like, I'm sure I read this in an old Krugman paper, I cannot find it, but if the primary appeal of a city or location is the network effect for a specific industry and/or general networking effects (like dating options in NY vs North Dakota) then increasing the supply can easily increase the value more than it increases the supply.
> This is a coordination problem: if every city upzones together, they can all get lower house prices, but each city can minimize its own prices by refusing to cooperate and hoping everyone else does the hard work.
There's a much worse problem involved: everyone who already owns property in a city doesn't want the property prices there to go down. "The city" might want that, whatever that might mean, but individual home owners don't. Nobody who bought a house for $1,000,000 wants it to cost $500,000 after the housing supply in the area doubles. All those NIMBY talking points about the crime and whatnot mask the real problem: they don't want to lose money because their house price decreases.
This is not a coordination problem and you got the incentives exactly backwards.
I don't see an ethical way of solving this problem, in a way that that isn't like, "I want to move into your city but the prices are too high so the people in your city should have half of their wealth wiped out so that I, not a resident currently, could move in".
That is, sure, building a megacity on the plains of South Dakota would work and then the property owners in other megacities complaining would look just as ethically unfounded as people complaining about NIMBYs. I think that it's a good idea.
You're missing the basic tradeoff of a city: the closer to the central business district, the higher the prices. The bigger the city in total population, the higher the price in the central business district, because there are more people fighting for that desirable place with the best transportation network access to everywhere else. This central area is expensive because of demand, but also because of physical constraints, skyscrapers are inherently more expensive than wood-frame buildings. The higher the skyscraper, the more costly it is. New housing will never make Manhattan affordable.
Where there is room for affordability is on the periphery and on transportation lines going into the central business district. This is the main lesson of Tokyo. One might look at Tokyo's central business district's prices and tiny expensive apartments and conclude market urbanism is a failure to achieve anything. But along the train lines that go into Tokyo, things get rapidly more affordable. YIMBYs should still support bigger skyscrapers in central business districts, but they should only look for affordability further out in 6 story buildings within walking distance of high-quality public transit lines.
So if NYC upzoned single family homes in the periphery burrows and improved transit, housing in Manhattan would probably get more expensive, even as there was increased affordability for the area from which one could commute into Manhattan. It's possible in the short run we might have uneven growth in demand that could continue to outstrip supply. Trains can move lots of people, cheap 6 story buildings can accommodate lots of people, and even if the ocean blocks expansion in one direction, that still leaves a half circle that grows the area people can live in quadratically. There's lots of capacity to meet demand to live within commuting distance of a desirable city center if there's political will to build transit and allow housing.
Edit: After reading other responses, I think I might agree. It's unreasonable for YIMBYs to expect small increases in housing supply to necessarily decrease short-run prices. Achieving affordability would be a long-run project. Short-run expectations should be heavily managed.
Density is not a destination or a characteristic of an area (and it's unfortunate we talk about it that way in Yimbyland).
- The more pedantically correct Yimby claim should be "increasing supply elasticity in the housing market, via regulatory reform, will reduce housing prices relative to local incomes and blunt price increases during economic expansions as well as set up loca housing markets for bigger price reductions during recessions (because we let developers build themselves off a cliff at the top of the market)."
Density doesn't drive prices.
- Prices drive density (within the constraints of the local land use regime), and prices are driven by demand. Yes, if we allowed denser construction, which definitionally would mean a reduction in per capita land consumption, ///relative/// prices would decline and more people could move in to a high demand area...but now we're talking about more people moving in to a high demand area ///because of the decline in relative prices///.
All cities are not perfect substitutes for each other for prospective residents.
- There surely some people doing the digital nomad thing or choosing which McKinsey office to work from in their late 20s, but I see no evidence that this population is large enough to have any appreciable impact on housing prices. Folks' personal and professional networks aren't perfectly distributed across all major cities.
What Scott is missing is that "density" isn't one thing. To a first approximation, it's comprised of JOBS and HOUSING. These are complements. If the amount of JOBS density increases, the price of HOUSING increases. If the amount of HOUSING density increases, holding JOBS constant, then the price of HOUSING falls.
This isn't at all counterintuitive and comes out of the canonical Alonso, Muth, Mills model (although without my cartoonish names).
Economics PhD student here, and first comment here.
I think Scott's basic framework is mostly right, but could use some tidying. His starting graphs basically depict a supply curve -- a positive relationship between prices and quantity, because if prices are higher then developers build more houses. But he's confusing a shift of demand along a supply curve with a shift of the supply curve itself.
Here is what I think Scott is trying to say. There is a simple supply and demand model that YIMBYs appeal to, with a simple result that pushing out the supply curve increases quantity and lowers prices, hooray. Scott is saying that there is a knock-on demand effect, what he calls "induced demand" and what activists call "gentrification" and what economists call "endogenous amenities".
This is a thing that definitely exists, but the relative sizes of the effects will determine what the end result is, and we also have to bear in mind that development occurs in response to changes in price. And the studies that Yglesias points to are essentially concerned with precisely disentangling and measuring these effects. The results generally say that the supply and demand curves are such that the price ends up lower overall, due to a combination of the slopes of the curves and the sizes of the shifts. We frequently do observe increasing prices and development together, but this is because the increase in prices causes development, which then (as the studies show) causes prices to be lower than if development had not occurred.
So that is why density does not cause price increases -- increases in price cause development, which causes a counterfactual reduction in price, which outweighs the effect of the endogenous amenity of density.
I'm not sure that the demand-increasing effect of density is big enough to make a substantial difference in practice. E.g. if lots of housing, enough to constitute a large fraction of the total, were built quickly in an already large city, then the opportunities arising from density would go up, & probably so would demand. But on the margin, the effect of each unit of housing on the opportunities arising from density is minuscule, so that that marginal upward effect on prices would probably be outweighed by the downward effect of normal supply & demand. (& of course, if a similarly enormous amount of housing were built quickly in a small city, it probably wouldn't cause demand to increase in the same way because there would be no guarantee that enough people would move there to generate the opportunities produced by greater density.)
My guess is that price as a function of housing supply would follow a relationship along the lines of Price=M*Quantity+B, where M is negative but B increases enough to offset that if the population density goes up by a lot (this depends on other things besides quantity supplied, like the local economy, so that New York, a typical small city, & the empty field in North Dakota can't be modeled as two points on the same line).
Good piece. As a longtime gentle YIMBY skeptic, I've always been a little bemused that the same crowd insisting that the rental market is 'just supply and demand' would be horrified if you said 'immigration obviously reduces wages, don't you know about supply and demand'. They would rush to explain to you that no, wages are much more complex than an Economics 101 understanding of supply & demand, that the new immigrants create demand for new goods and services and so actually create more jobs by their presence, and so on. I would personally agree with this pro-immigration argument, I just think it's funny that to the YIMBY crowd simplistic 'supply & demand' explains real estate prices but doesn't explain wages.
Cities are huge agglomerations of network effects- they're valuable precisely because there are more potential employers, romantic partners, friends, business connections, activities and more- it's quite possible that a higher population makes the city more valuable and so raises rents! (And wages). I think Devon Zuegel had a post about this from a few years ago. Anyways, 101ism is a bad way to understand complex systems
I think the basic problem is that you're ignoring other covariates in your regression, most importantly the demand to live in a city. As a result, the regression coefficient is incorrect. This is a well-known problem in statistics, called the omitted variable bias (https://en.wikipedia.org/wiki/Omitted-variable_bias). I believe if you account for the other variables then the regression coefficient will be negative as expected from economic reasoning.
This is why marginal analysis is important in economics. While comparing two situations many variables will be different, so it isn't clear exactly how the outcome is related to one variable in particular. In marginal analysis the two cases are pretty much the same and the only difference is the marginal change in one variable. So any change in the outcome can be attributed to that variable.
I think the answer to this hugely depends on your internal mental flowchart. This post seems to suggest a model of:
housing -> cool place -> high prices
Which I think buries the step that is:
housing -> support for cool amenities -> cool plate -> high prices
This in turn allows for some finer grained analyses: If Norman, OK (the lowest population density city in the US) built even a moderately large amount of housing, it probably would crash housing market there - no matter how much housing there is, it's not going to experience a housing boom. On the other hand, Austin TX strikes me as too far down the path of becoming a "cool city" to avoid price growth by building housing. The marginal person probably avoids moving to Austin at this point because it's too expensive for the existing amenities, rather than because there's not enough to do there. And finally a city like NYC or SF is probably already "too cool" to realistically discourage interest by preventing construction.
Oakland then becomes a bit of a weird case - it's close enough to a major high cost area that if it got cooler it really might start growing new amenities (restaurants, better (perception of?) public safety, better public transit, more people you know, etc.) and the increased amenities might in turn drive increased demand. But it's not really cool enough that its growth is locked in, at least as far as I can tell.
I think the "ground truth" here is probably somewhere between your and Matt's position - construction really seems to have a local, short term, effect in raising prices; construction is a key element in becoming a successful big city, which is correlated with higher prices; and the exact trajectory of a city between "low cost, mid sized city" and "high cost megapolis" is probably determined by some amount of coordination problems that are fairly opque.
I don't think it's obvious that there is any agglomeration effect from the data presented.
Yes, there are some Americans who want to live in a big city. This is part of what makes New York more attractive than Lubbock, TX for a random 30 year old. But your charts make a comparison of density within a class of "big US cities". Do you really think people consider Houston and SF, and think "yeah I'm going to live in SF because it's really dense. I know it's half the population, I just want the density." It doesn't pass the common-sense test.
The reason people move to SF and NY is because of jobs and culture, not specifically how scarce housing is. In fact, that is an obvious reason people don't move to those cities. The culture/jobs demand drives up house prices, and consequently makes cities more dense. But if you just build more houses, quantity supplied is rising due to strong demand. There's nothing magic about density itself. If you build more houses, you have a shift right in supply. This lowers prices.
It's a fair argument to say, "well why hasn't this worked anywhere in America?" I guess the argument would be, it hasn't really been done on a sufficient scale, so any supply-side effects are overwhelmed by demand-side effects. But I'm not confident in this. Just not convinced that building houses has any positive causal effect on local house prices.
> They moving there because they want to be in a big city - with friends, jobs, museums, and nightlife.
Ok, but these things are not synonymous with density. In fact, having more office space, museum space, restaurant and nightclub space, etc. means having *lower* population density, ceteris paribus, because you can't live in any of those other types of building. People would still want these things if there were no big dense cities. Either they wouldn't exist, or people would still try to live in the places where those things are available, which be more expensive than NYC is today. Basically, I think that the effect of density -> amenities is much smaller than the other way around. Cities are built for economic reasons--you say " Manhattan has a few extra natural amenities, like a river and a good harbor. But nobody moves to Manhattan for the harbor. They moving there because they want to be in a big city - with friends, jobs, museums, and nightlife." but this isn't really true. The harbor is why the city exists at this location in the first place. It's no longer the direct reason why NYC is an economic hub, but its current major industries (finance, research, media, tech, etc.) are there in large part because it was previously a center of immigration, trade, and manufacturing.
Building more housing would not turn Oakland into SF or NYC.
Consider also that the population of Manhattan peaked around *World War 1* (https://en.wikipedia.org/wiki/Demographics_of_Manhattan) at around *40%* more than today. Do you think it was cheaper to live in back then, or more expensive? Given the average salary of a factory or port worker, I'm going to guess cheaper.
>. But doesn’t induced demand violate the economic law of supply and demand?
No. See https://www.reddit.com/r/badeconomics/comments/8ghb8j/r1_what_is_induced_demand/. (It's also related to https://en.wikipedia.org/wiki/Jevons_paradox).
> if every city upzones together, they can all get lower house prices, but each city can minimize its own prices by refusing to cooperate and hoping everyone else does the hard work.
I don't think this is what the data really show. At this point, building housing in the expensive areas would just serve the already existing amenities that are already drawing people there. And it seems like it's a fairly common development pattern in the US for suburbs to be more expensive than the city they surround. But also, many towns don't *want* their own housing prices to be cheap. "Diminished home values" is one of the most common NIMBY talking points against, well, everything. If you own a home, then expensive housing is good for you, and bad for everyone else.
Ok, really silly. We are comparing the overall "life opportunities" of North Dakota and NYC and then wondering why housing is cheap & low density in N Dakota? Really? Maybe application of common sense should come first before writing.
"One common pattern is to prefer any big city"
Proof that this pattern exists is?
https://en.wikipedia.org/wiki/List_of_United_States_cities_by_population
Sorted by population density, NY #1, Jersey City New Jersey is #2, and Paterson New Jersey is #3.
Newark is in the top 10, and Philadelphia and Yonkers are in the top 15.
So your thesis appears solidly disproven.
People either want to live in city X, or they want to live in a small set of cities Y. They do not want to live in "any city Z with sufficient population density"
Rather than disagreeing with the conclusion, I want to object to the whole framing of focusing on the effects on housing prices.
It’s not immediately clear what police goals should be with respect to home prices. Existing homeowners who think of their homes as investments would like prices to go up; renters and new homebuyers want prices to go down. Numbers on both sides of this issue are close enough that neither position is a political slam-dunk, and nobody on either side is going to change their mind easily (unless they buy a house I guess), because they’re all voting in their own economic interests.
So the end result of focusing on home values is always going to be political paralysis, and framing the issue in those terms already gives the game away to NIMBY forces that profit from political paralysis.
I don't have strong opinions on the details, but I would be cautious about false assumptions of linearity. Part of what makes places like Manhattan and London expensive is the network effect, which makes them desirable. Density is one variable that contributes to their network effect, but:
1) it is not the only such variable, and
2) even only considering density, network desirability is not a linear function of density.
It doesn't seem crazy to me that, due to other variables, add a little bit of housing, in a way that doesn't much change network effects, could move prices in the opposite direction from adding a great deal of housing, that did.
"I don't think there are enough immigrants to really affect things."
Given that 31-39% of the Bay is foreign-born _right now_ and the global population of mobile individuals ready to leave their homes to work for a wage lower than prevailing Bay Area wages is at least a billion, maybe more... how can it not affect things? Net immigration is in the millions per year to the US... all of them by definition need somewhere to live... how the Bay vs. the US. vs. the World a self-contained proposition?
Source on the 31/39...
American Community Survey (ACS): https://www.census.gov/programs-surveys/acs
Select "Foreign Born" and then the "geography" filter to either San Francisco-Oakland-Berkely (31%) or San Jose-Sunnyvale-Santa Clara (39%)
My fundamental objection to what you're saying is you're just looking at correlations and reaching a kind of "ice cream sales cause murder" conclusion (referencing https://www.lifehack.org/624604/the-most-common-bias-people-have-that-leads-to-wrong-decisions).
I think we might agree to a rough model of "Demand = Price + Supply", and lets first consider what happens when demand is fixed: high demand expresses in high price and supply. Randomly generating some cities with high and low demand, you'll see high price and supply co-occur, as in your correlations. But once these cities are made and their demand is fixed, a hypothetical urban planner sees price and supply trade off each other, so increasing supply is how you decrease prices.
Now obviously demand is not fixed, and supply up -> price down -> demand up could happen (and I would wager does)! But what is the coefficient? If X>1 and each new home makes more than one person want to additionally live there, then building new homes raises prices, while for X<1 econ 101 holds and increasing supply decreases demand. This is precisely where you need to look at the empirical literature to measure X, and it says upzoning lowers costs.
I'm looking out my window at the crowds gathered to catch a glimpse of the celebrities arriving at the Met Gala in a few hours.
No matter the number of new housing units built, would Oakland ever have something like the Met Gala or in the case of LA, the Oscars? Maybe in a hundred years. It takes a great deal of time and luck to become a New York or an LA.
As for Miami, it's a contender, but way too early to tell.
1. This is why I'm so excited about the recent YIMBY bills passed at the state level in California. I think a change at the state level is the only way to move forward instead of hoping that each NIMBY city changes their minds individually.
2. This argument contradicts the stereotypical NIMBY reasoning of "if you build that apartment building 2 blocks away, it'll add more supply and the value of my home will go down". They should support housing, because the value of their property might actually increase!
3. If there's really people out there who are willing to choose "any big city", shouldn't these people move to the other cities as prices keep going up and up in NYC and SF? Maybe people and companies will start moving to Austin, Miami, Atlanta, Cleveland, Detroit, Sacramento, Minneapolis, etc. Honestly SF and NYC are not worth their high price *right now* in my opinion.
I think you are in general right that agglomeration effects are real, which is why bigger cities have higher value to residents. I agree that people move locations. But I think you can go a step further. If one city is growing faster and densifying, surely those people are not demanding homes in other cities and those cities build slower. This is part of the spatial equilibrium story that further makes claims about “build density and get cheap homes” less plausible.
My background: I used to work for property developers and have been researching housing supply for many years https://scholar.google.com.au/citations?user=P5s-_d0AAAAJ&hl=en&authuser=1
It is not reverse causation but rather the existence of a confounder - the attractiveness of the location. Imagine cloning that island off the coast of Maine and gluing it to Manhattan. Many people will want to move to the clone, but few will want to live on the Maine island. 20 years from now, after developers have built housing on the NY island, it will be both denser and more expensive than the Maine island.
Also, don't forget that demand is the willingness AND ability to buy. Those dense cities also have lots of well paid people. That increases demand and drives up prices. You need to control for income before concluding that denser places are more expensive.
Edit: Point 3: I just noticed that the Y axis is pct of income spent on housing. That will also be affected by income. Someone who takes home 10K after taxes will struggle if he pays a third of his income for housing. Someone who takes home 50K can afford to spend half of it on housing and still have plenty left over.
Popularity breeds popularity; that's a "network effect," and it holds for cities just like for music concerts or second languages.
But does popularity breed prices? That depends. Scalped tickets to hot concerts are super-expensive; learning English isn't.
Both "tickets to the popular concert" and "English proficiency" have network effects, but only one has limited quantities available. The price effect is obvious.
American cities, and most cities worldwide, limit the quantity of housing. So in most cities, popularity acts like concert tickets: density goes with high prices.
A very few cities like Tokyo don't limit the quantity of housing. So there, popularity acts like English proficiency: density goes with flat prices.
I think there are both amenity and disamenity effects from density, but the real thing that pushes up housing costs is having lots of jobs nearby. To some extent, jobs will tend to cluster near housing, but both job decentralisation in the post-war era and re-centralisation since the 90s show it mostly happens independently of housing growth in this our very regulated market.
I think this is missing something by categorizing a city as entirely "lots of people (and their housing)." You can do better by modeling cities as containing 2 categories of stuff: housing and attractions. People want to move to a city because of attractions. Building more housing raises the number of people able to live in a city without increasing the number of people who *want* to live there--possibly, in fact, decreasing it (because people want to live in areas with high attactions per (area, resident, etc)) In this model, there are 4 different ways you can move--increase housing, decrease housing, increase attractions, decrease attractions--each with obvious effects on pricing.
So, really, if we want cheaper housing in (say) NYC, we should bulldoze central park, close all the museums and restaurants, and ban broadway plays.
I am so excited that you've finally thought yourself almost all the way around to becoming a NIMBY like me.
The final bit you're missing, I think, is that the best solution is to build a bunch more housing everywhere _except_ the biggest metros. The San Francisco Bay Area has eight million people, which is too many, and its geography is amazingly poorly suited for a giant metro area (what with the huge bay in the middle, and the steep hills, and the earthquakes).
We need to find ways to stimulate growth in cities in the 100K-1M range, and maybe the 1M-3M range, without cramming more people into the small number of overcrowded megalopolises that already exist.
It matters what is being built. People don’t move to NYC because there are so many houses there. It is the industry and culture. If you are building more things like stock exchanges and museums, housing prices will go up at a constant housing density. That doesn’t mean building more houses will make housing prices go up.
Your argument works specifically because you assume building a denser and more desirable city.
You could make the housing prices go down by building a denser and more horrible to live in city, if housing prices are that important.
I think the unstated assumption is that size in and of itself is what attracts more people to a city and hence increases demand and prices, rather than something unique about the cities, like architecture or neighborhoods. It’s not intuitive to me that New York sized Houston becomes popular just because it’s big. (This hints that other features that made cities desirable started the flywheel.)
The "induced demand" argument essentially says that there are a significant amount of people who don't want to live in area X now, but would if it were a bigger city. I can kind of see the argument for this in some cities, but who could possibly be thinking this about NYC and the Bay Area? Those are already *the* places to be - building more housing can't make them any more so.
Also, I'm going to nitpick the density-vs-pricing graphs you picked. The first one is weighted by a bunch of unlabeled data points on the lower end, and the second one shows a frankly unimpressive R-squared value.
Off topic I know but I can not find the SSC post talking about how scarcity leads to politics and scott makes up a fictional example of if there was 10% of the available water all the interest groups would fight for it but with enough water it is a nonissue. It could be a ACT post but I think its older.
I think the hypothesis isn't clearly defined, and the question is phrased in a way that makes a correct answer impossible. If building more houses causes local prices to go down for 4 years, and to go up from then on, did it make house prices go down, or up? If building more houses causes others to build more amenities, and attracts more businesses, creating more jobs... how far down that chain can we go, and still attribute raise in local house prices to building more houses?
Each of those additional steps--building museums, creating new businesses, etc.--costs additional money. It seems to me that the credit for the rise in housing prices should be distributed among the different factors proportionally to their capital (or other) investment.
You're approaching the question like a foundationalist, who believes that assigning credit for a result to a single cause is unproblematic. Correct epistemology must allow the dynamics of recurrent causal networks to be computed iteratively, and an answer to be given in a form that could reproduce dynamics like "price goes down for 4 years and then goes up".
It might be interesting to look at the Spanish market for a possible counterpoint.
Spain has a very uneven population, with most concentrated in the coast and the center. The sparse parts of the country are very sparse, and the dense ones are very dense.
In the rally to 2008 Spain was building A LOT of new units [1]. Prices were rising at a very rapid rate. At the same time population was growing fast, mainly due to immigration [2].
When the bubble burst in 2008, home prices started going down[3]. Not surprising considering the inflated prices and the massive uptick in unemployment [4]. However what's surprising to me is that this price slump lasted a really long time, and only recently we are approaching prices close to the top of the bubble. Is is true that population growth slowed down a lot, and for a few years between 2013 and 2016 total population actually went down. However median salaries [5] sort of grew steadily over the period, again, not accounting for unemployment. Still, the bottom of prices seems to have happened around 2016, when unemployment had been decreasing from it's peak for a few years.
Of course there are multiple interrelated causes happening at once in this case, but my take is that high density helped to tame housing price for a relatively long time, even as economy recovered from the recession.
[1] https://www.eleconomista.es/empresas-finanzas/noticias/10318431/01/20/Se-construyen-muchas-o-pocas-viviendas-en-Espana.html
[2] https://es.wikipedia.org/wiki/Demograf%C3%ADa_de_Espa%C3%B1a
[3] https://www.idealista.com/sala-de-prensa/informes-precio-vivienda/
[4] https://elpais.com/economia/2016/04/28/actualidad/1461866236_955613.html
[5] https://www.newtral.es/sueldo-medio-espana-evolucion/20220427/
I think a lot of the studies showing building reduces costs locally are accurate in the short term, but the reason you're right is that it takes time for the new supply to attract out-of-area residents, increasing the population, and increasing desirability. A lot of this can be a multi-year cycle, i.e. New York builds some housing, on the margin your friends want to move there, they do, 3 years later a critical mass of them is there, and you move there as well. And then at this point, it actually did contribute to a raising price.
I think what you and Matt are both saying could be consistent. For example, it might well be the case that on the margin adding more housing to New York doesn't increase its "New Yorkiness", i.e. its having of Broadway, Wall Street, and so on, because those are features of being the #1 city, not features of being a city of N million people. In this model there could be discontinuities when you get enough extra people to cause some sort of runaway migration to your city, but it wouldn't increase demand more than supply on the margin.
Of course this could totally be wrong, and it could be that with 5% more people New York would _also_ become the center of software in the US or something.
A topic close to my heart - and I suppose many other's too. Thanks for this. I am EXTREMELY skeptical of any claim that building more homes lowers prices or eases affordability for families. I live in an area that built tens of thousands of new units and rents in these buildings basically monotonically increase.
There are a few main factors as I see it (well, maybe it's really just one):
1. Living in a globalized world. More houses for sale is just more inventory from which wealthy domestic and international buyers can select from. If there is an ephemeral drop in price, it will entice more of these buyers until the situation returns to exactly as it was before.
2. Everyone in the housing market (buyers and sellers) have wisened up and are playing the same game. With the explosion of equities and real-estate following the zero-interest rate environment for so long, people's net worth blossomed. Though equities are down a little, most have now wisened up and have moved that wealth into money market and related bond funds that are earning them a nice 5-6% interest rate. This means lots of people have lots of money - in my area development just keeps going further out and, even in April 2023, open houses are packed and houses sell after 2 days for 100s of K over asking - often in cash. Just as before, if creation of new housing temporarily lowers price due to a higher supply, it invites an even larger horde of potential buyers who are in the (n-1)th tier of wealth - and then it returns to the equilibrium of before, where everything beyond the affordability of almost everyone, yet any house on the market (unless utterly dilapidated). Building more houses just invites more buyers from more places -- and rinse and repeat.
3. Building new housing (particularly in wealthy, economically dynamic areas) increases the population and with that comes more business, jobs, investment and all the things that make a place desirable and expensive. The ultimate attractor of the system is to be unaffordable. If it was affordable, everyone would buy it all up until it was unaffordable all over again. I think the logic is as simple as that. Maybe a wealth tax? And/or a severe recession where people lose jobs and wealth and HAVE to sell are the only things that would change this.
I think the YIMBY argument is that it's interesting that altering the amount of housing in an area, resulting in higher density in that area, doesn't result in higher prices. You seem to be saying "ah yes, but under our current planning laws, more people moving to a city does increase prices in that city".
This seems like the same argument seen from different angles. You're making the point that in the past, prices rose with density on a city-wide basis, even if that doesn't hold for small areas within a city. Which sounds like something YIMBYs would agree with.
I think the response might be "yes, the point of the small scale studies is that they suggest it's not increased density itself that increases prices. It's something else, something that only shows up at a larger scale, over a longer time period, and we think that that something else is the overarching planning regime".
And that's where Tokyo comes in - you can say, look, there's another wealthy, developed country with far less restrictive planning laws, and in that country increasing density doesn't seem to lead to increasing prices.
If you've got a vetocracy, increasing the number of people with vetos will probably decrease the number of laws passed. But that doesn't mean that it is inevitable that larger populations are more inclined to paralysis. It means you have to change your system of government.
I'm skeptical about this model at the city level but I think there's some truth to it at the micro-neighborhood level. Building ten new homes on my block would not meaningfully increase affordability of my immediate area, since this block is more or less fungible with dozens or hundreds of other blocks surrounding it. If anything, it makes the micro-neighborhood slightly "fresher" and should increase property values in the immediate area (for example, because those existing properties are now adjacent to brand new homes instead of overgrown vacant lots or crumbling warehouses). This is the core of the NIMBY problem in my view: building housing on my block has diffuse affordability benefits far beyond my block but very concentrated harms (living through construction, etc).
According to AAA (old data from three years ago) the cost to own a car in America is about $800/month, including depreciation, insurance, gas etc. So living in a community where you can take transit or a ride share easily can make up a significant portion of higher rent. Some. Not all. Just something to keep in mind in moderate priced regions.
Re: SF; a few years ago I saw a study where the researchers kept drilling down on why NIMBY's hated new development, and the biggest draw was hatred of already rich developers getting even more rich. So the green monster called jealousy may be a factor in a classism response.
The driver that both sides seem to be missing — or at least are not being explicit about — is that _density is necessary but not sufficient for valuable things beyond more housing_.
In particular, density allows for thick labor markets and increased specialization, making people more productive; and thick social markets, making life more enjoyable (try finding a date today to take to a show tomorrow in NYC vs. SF!). Those things take time for cities to develop and are dependent on things other than density, but they are valuable, and so folks will pay for them if they come to exist years down the line.
That's why:
- building more housing in NYC will immediately decrease rental prices
- building a huge amount of housing in rural China won't increase rental prices alone
- ...but building a huge city in rural China _and_ having an industry cluster form around it — such that people have productive work and enjoyable lives — _will_ increase rental prices
- inner city Detroit is both dense and extremely cheap, and building more housing won't change that
To believe that building more housing in NYC would increase rental prices in five years, we need to believe that more density makes living in NYC _even_ more valuable. Is that the case? I'm not sure — maybe beyond a certain point of density, it doesn't make us more productive or our lives more enjoyable. Or maybe there's no limit.
Regardless, the model of "Density Increases Local But Decreases Global Prices" is unambiguously wrong for places that can't build those other things — inner-city Detroit as an example.
>The two densest US cities, ie the cities with the greatest housing supply per square kilometer, are New York City and San Francisco. These are also the 1st and 3rd most expensive cities in the US.
Let's distinguish between the housing stock (quantity of homes) and the supply curve, which represents the quantity of homes as a function of price. Also note that price is determined by the intersection of the supply and demand curves.
https://en.wikipedia.org/wiki/Supply_and_demand
>For example, if my home city of Oakland (population 500,000) became ten times denser, it would build 4.5 million new units and end up about as dense as Manhattan or London.
Are you thinking of a shift in the supply curve (moving along the demand curve), or a shift in the demand curve (moving along the supply curve)? If you shift supply, prices go down and quantity goes up; if you shift demand, prices go up and quantity goes up.
>Right now, more Americans prefer to live in big cities than there are housing units in big cities, so prices go up and these people can’t afford their dream. As new cities become “big” (by these people’s criteria), they’ll move to those cities, increasing demand.
Assume a two-sector supply and demand model, where we have graphs for Big cities and Small cities.
In the initial equilibrium, prices in Big cities are higher, because more people want to live there than there are homes. But note that Small city people still show up in the demand curve for Big cities: they're in the section below the equilibrium price, because their willingness-to-pay is low.
Now suppose that a Small city (somehow) transforms into a Big one. This shifts supply right in the Big city market, reducing price. If the price reduction is large enough, we move down the Big city demand curve so that some Small city people are now willing to pay the lower Big city market price. Note that Big city demand hasn't increased; it was always there.
(Exercise for the reader: what happens in the Small city market?)
At the micro-economic level, adding more housing units can indeed "lower" the median household price in a given metro area. Even in an affordable city like New Orleans while simultaneously increasing the adjacent cost of housing in certain neighborhoods via market stimulation. The best example I can give is an affordable housing project I worked on that was comprised of 79 affordable units for working artists (income qualified). It was renovating an abandoned school in the historically black neighborhood of Treme. The school had been left abandoned since Katrina (2005) and the immediate surrounding blocks reflected the slow decline of the nieghborhood hollowed out by dis-investment, subsequent broken-window crime and poverty.
Shortly after our project began construction, local builders and individuals began buying adjacent abandoned properties and either renovating/constructing new single and 2-family homes. By the time our project had been completed, the 79 units helped to lower the "overall" median rental price yet the adjacent neighborhood saw an increase in housing costs as gentrification took hold. This story repeats itself in other cities I've worked in like Denver, Washington D.C., Nashville, and Dallas. Phoenix was the outlier in that there's so much sprawl, gentrification can't really happen.
I have yet to see an in-depth study that quantifies the impacts of new housing units on market pricing at the overall median price level for a city and the immediate local affects of those same units on a particular neighborhood. I think the NIMBYs and YIMBYs talk past each other in this regard where each point to the relevant data that best supports their position without taking into account the complexity of how increased housing density affects are regional/metro and hyper-local. Which is why you have YIMBYs like Matt claiming more housing ultimately lowers the median housing price (nominally/briefly) while NIMBYs claim every new neighborhood development causes housing prices to rise and gentrification by yuppies/hipsters/bohos push out the the BIPOCLGBTQ+ artist/teacher/essential oil craftsbeing/zen gardner/retiree.
The conclusions you draw from your urban density & housing affordability correlation graphs are falling prey to the classic correlation is not causation problem. You acknowledge the potential reverse causation effect, but there are other reasons correlations don't imply causal relationships that you are missing from this analysis. In particular, you are not accounting for an omitted variable that affects both urban density and housing affordability: desire to live in e.g. Oakland vs. Manhattan, AKA housing demand. You correctly identify that induced demand is a thing that sometimes happens, but induced demand does not describe the whole dynamics of density and housing prices--you yourself say: “[People are] moving [to large cities] because they want to be in a big city - with friends, jobs, museums, and nightlife.” These are related but not perfectly correlated with density of housing for a multitude of reasons. Not all big cities are interchangeable.
Or: there are three relevant factors to the theoretical YIMBY argument about how marginally building more houses lowers housing prices, and you are kind of ignoring how demand/desire to live in a specific place is a separate thing from density, and should be controlled for when you graph a price function of density. This omitted variable biases your causal conclusions.
At least, that’s my read. If I got any of your arguments/thinking wrong, I do apologize.
This paragraph is wrong I think:
“For example, if my home city of Oakland (population 500,000) became ten times denser, it would build 4.5 million new units and end up about as dense as Manhattan or London. But Manhattan and London have the highest house prices in their respective countries, primarily because of their density and the opportunities density provides. I don’t see why Oakland being able to tell a different story of how it reached Manhattan/London density levels (“it was because we were YIMBYs and deliberately cultivated density to lower prices”) would make the end result any different from the real Manhattan or London. But if becoming just as big as Manhattan or London would make Oakland more expensive, shouldn’t we assume that a little step in that direction would make it a little bit more expensive”
If Oakland all of a sudden had 10x the housing supply I would NOT think it would be as expensive as New York. I think you would have a ton of empty housing and princes would be dirt cheap as people tried to get something for their homes for which there was not enough demand...
The reason we don’t see his is that no developer is going to build a ton of supply beyond demand. Since it’s so hard to build this is rarely the problem. But we see it when cities de-populate. There was a time when Detroit homes in some areas were going for ~free. If someone added a bunch more supply in Detroit it would be unlikely to “induce more demand than supply”
It seems like a lot of the argument rests on “we can all agree it is true at the extremes, so doesn’t that counter the studies that show it is not true at the micro level”. But if someone (me!) doesn’t believe the extreme argument then the studies that show that induced demand does not outweigh increased supply effects seem totally plausible.
So I disagree with this argument for a lot of reasons.
I think the first (and easiest) counterargument is a money-on-the-table one. Let's suppose you're right; that if a town has a density similar to Athens, GA (where I live, good sized town, 100Kish people), then it should have housing prices similar to Athens (about 300K for a single-family home). Athens has lots of ordinances about what can and cannot be built, you are going to have to apply for permits, etc (this is true anywhere). So, just buy up 100 square miles of land in rural Georgia and build a similar sized amount of housing. If you did this, your profits should be astronomical, since you would have very little regulation to fight (this is especially true of building an Oakland sized city in not Oakland). But we don't see venture firms actually building brand new cities wholesale in America, so maybe this should tell us that there actually isn't money on the table here.
2: I think that this has been mentioned before, but jobs are probably a bigger driver of rent. I think this is especially true in single-employer cities. If you consider most rust-belt towns, their land value dropped like a rock not because more/less houses were built, but because the factories closed and were automated/went abroad (this happened to my hometown). You can even see this in bigger cities like Rochester, NY, which has a pretty high population density for a city, but not an extraordinary cost of living. It used to have Kodak as its #1 employer, but when Kodak closed, the city took a huge hit. Maybe I'm wrong about the numbers, I'm open to someone beating me back with hard data, but this is my overall impression that a lot of cities/towns lose land value from loss of employers, not housing issues.
3. I think that there's an important replacement rate issue. I do think that you're right in general, that as a city grows, housing becomes more expensive. But what I think is going to matter is how quickly housing grows relative to population; if a city could maintain 10% of its occupations as uninhabited, then I think it's land prices are going to be lower than a city with 1% uninhabited, even if their populations grow in lockstep. I think though that as a city gets bigger, its regulations are going to make it harder and harder to add new housing, and so the percent of uninhabited places is going to slowly drop, adding to the rent increases.
There is also a question of why any property owner would build so quickly that it forced prices down. There is an inter-temporal equilibrium here too as I explain in this paper.
https://link.springer.com/article/10.1007/s11146-020-09815-z
Many people simply assume that changing planning rules will change how fast new building happens. Sure, it will probably change where it happens, but probably not how fast more housing is built across all substitute locations.
i think it would be helpful think in terms of elasticities rather than in levels. wine (nyc/sf) is more expensive than coffee (north dakota). but if demand for either increases, price responds very little because supply is inelastic. if more people want coffee, coffee producers just harvest a few more beans, hire a bit more labour, etc. marginal cost barely increases, and if the market is competitive same for the price. if housing supply were elastic, there would still be a gap between nyc/sf and north dakota (like the gap between wine/coffee), but it would be much less. the current difference reflects inherent factors (better amenities and higher real wages in nyc/sf) and the "distortion" of inelastic supply causing excess demand to drive a wedge between prices
the point about cities being in competition is just another way of saying demand is elastic (since cities are close substitutes for each other), so any increase in supply in a city won't lower own-city prices much. this isn't a problem if the people deciding on how many houses to build are "price-takers" (i.e., they don't consider the effect of their housing construction on overall prices). as long as MB > MC at current prices, they'll build and the market will clear. the current situation--where interest groups who *want* prices to rise use regulation to make supply inelastic--is the opposite. it's more like local political economy oligopsony than classical general equilibrium perfect competition.
i also like to think of it in terms of flows rather than stocks. suppose there are 1m houses in your city (stock). mostly people occupy their houses year after year. every year, 10k people show up in your city looking for a house. if there are fewer than 10k vacant houses V, they bid up prices. if there are more than 10k vacant houses, prices drop. it's the ratio of vacancies to people looking to search V/10k (what economists call "tightness") that determines the price (i.e., it's the 10k flow not the 1m stock)
if a city had a one-time extreme increase in its housing stock, but then implemented laws to make the future housing stock inelastic (i.e. the history of ny/sf), you would see low prices for a few decades, then over time as demand increases eventually prices would explode. this makes me optimistic because it means we don't have to make huge permanent investments in the housing stock, we just need to loosen things up enough so that every year, the number of houses coming onto the market exceeds the number of people looking for houses. this seems more manageable
Increasing supply has 2 effects (abstracting from the land price vs building cost distinction for now):
1) For any negatively sloping demand curve, it lowers prices.
2) Increasing density makes the city a more desirable place to live and increases wages through agglomeration effects. This shifts the demand curve for living in that city upwards/to the right, increasing prices
I think you're making the case that the positive correlation b/w density and prices tells you that effect 2) usually dominates and is what would likely dominate in Oakland if the Oakland were to build more housing.
I don't think this inference works because many amenities that make a city desirable aren't perfectly correlated with/entirely caused by higher density. That is, dense regions can be expensive for two reasons:
A) They happened to have nice amenities for reasons that are not caused by their density (e.g. nice beaches nearby). These nice amenities increase demand and prices.
B) Density causes better jobs and better amenities as in point 2) above.
A) would cause the correlation between density and prices to be positive (if weaker) even absent B).
For an example of A), were NYC density to reduce by 20%, Broadway wouldn't decrease in quality by 20%. Similarly, if Miami decreased it's population by 20%, I don't think it's amenities (the beaches) get 20% worse.
Similarly, if Oakland increased density by 20%, it becomes a moderately more desirable place to live because of B). However, if A) explains a substantial share of the variation in desirability, then Oakland doesn't attain the amenity/desirability/price level of a city that currently has 20% more density than it.
I think this means that it's totally plausible (though not guaranteed) that effect 1) is larger than effect 2) and that an increase in supply lowers prices. Indeed, this is what Matt's studies seem to show.
The missing link here is that by just focusing on housing prices, we are ignoring the other key components that go into what make people better off (what economists usually define as "utility"). People don't just isolate one component when making a decision, they consider the bundle as a whole.
Let's say that people's well-being has only two components: 1) quality of local amenities and 2) budget for non-housing consumption (think especially of tradable goods like iPhones and airfare).
It can make sense to move to NY from a lower cost city if you value the local amenities highly (walkability, Broadway), even if you keep the same income. You can consume fewer iPhones but you're still better off.
Now let's throw another important wrinkle in here: income. Big cities literally make some industries more productive (knowledge spillovers, scale), and some component of that productivity is captured by workers in higher wages. If moving to NY changes your after tax income from $50k to $100k, you can change your housing expense from 20% to 50% of your income and still have a larger budget for non-housing expenses ($40k to $50k) in this example.
The point is that narrowly focusing on minimizing housing prices is a dead end, because the other components of utility will tend to overwhelm it.
The YIMBY argument should be that construction lowers housing prices *holding everything else constant*, which increases utility - more people living in high productivity areas (more economic output for everyone) and having access to amenities that require density (if that's what they desire, and people wouldn't move to bigger cities voluntarily if they didn't want it).
If building more housing in a city would increase the value of real estate in that city, perhaps we can count on political forces in that city causing more such housing to be built. (Renters might not like it, but they probably don't have nearly as much influence over the housing approval process as land developers.) Consequently, cities that are not seeking to increase the quantity of housing are likely cities where building more housing would lower the price of housing.
Why don't they build twenty-story skyscrapers in rural areas?
There’s also the question of whether cities can actually generate a sharp increase in housing supply from the usual policies that YIMBYs advocate, such as mass upzoning. There are good reasons to be skeptical. Recent work by Cameron Murray has explained why “landbanking” is so common and why rates of supply in cities are closely tied to market demand conditions, not to zoning policies. You should check out his research:
https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3417494
https://www.fresheconomicthinking.com/p/why-i-am-anti-anti-zoning?utm_source=%2Fsearch%2Fanti&utm_medium=reader2
There's a relatively clean theoretical answer, which is that demand is a function of social features (like good food, cultural events, ability to find your niche, variety of jobs...), and social features are a *superlinear function of population.* Social stuff is dependent on the number of possible pairwise interactions.. city X with population double that of Y will produce more than twice the number of patents, creative collaborations, etc... as well as more than twice the amount of crime, all things equal.
If social features scale superlinearly (and demand depends on it) and housing scales linearly, then you'll see housing prices go up as density increases. If you add housing without adding density, say by building outwards, you won't see the same sort of price increases.
But that's not all... what we should really look at is density and point to point travel time. A city that spreads out with good travel infrastructure may have housing prices similar to a city that builds up with mediocre travel infrastructure. I'd guess that population, how many other people you can reach within a fixed unit of time, and housing stock together predict most housing prices.
This analysis seems to have trouble explaining why anyone would ever build a detached suburban-style development in, say, Houston; all new construction everywhere should look as much like Kowloon Walled City as it is legally and architecturally possible to do so, because from the developer's perspective, density is a free money-making machine. In fact we basically never see anything like this, suggesting that short of maybe some Neom-like scheme on a truly grandiose scale, it is impossible for agglomeration-effect-induced-demand from normal building projects to come anywhere close to outweighing the supply increase.
What is the upper limit on geographically mobile yuppies ?
China is able to erect new ghost cities for a million+ people in a decade. If yuppies flood to a town, then there are no technological barriers to building at a pace of 100k/yr. I can't imagine 100k yuppies moving to Oakland every year. Inevitably, there will be a surplus and prices will go down.
The rise in local prices only persists when supply is limited. In fact, the local increase you speak off, is no different than scalpers re-selling new consoles at extortionate prices. This only lasts until manufacturers can get all their supply chains ready for the massive demand. Eventually, the manufacturers catch up, and the massive demand allows them to build their product at much lower costs due to scaling benefits.
With the deterioration of mid-western cities and their terrible weather, I'm surprised that internal migration to fair-weather states is as low as it is.
Everyone keeps wanting to bring up Tokyo, but they're ignoring the third biggest city in the US: Chicago. Chicago has world-class urban amenities and a broad range of high-paying employers. But housing prices are significantly lower there than in comparable US cities because they've consistently made it easier to build new housing.
Of course, you can bring up other reasons why Chicago might be less attractive, like Illinois politics and taxation, cold weather, crime, etc., but I don't think those are enough to explain the huge gap in housing costs. This kind of disproves Scott's model that there's a huge pool of people in the US who just want to live in a big city and growing bigger might attract them. If they were really as location-agnostic as Scott claims, they'd all be living in Chicago now. (Or maybe Dallas or Houston, the fourth and fifth largest metro areas in the country, that again have significantly lower housing costs).
So according to your hypothesis, homeowners should support construction of new housing nearby, because it will increase prices of the homes that they own. But typically we see the opposite. Are homeowners wrong?
Anyone interested in this topic should read the paper from Moretti and Hsieh on it. Examines the effects of reducing regulation for land use and how it’s related to employment, wages, and city size.
https://www.aeaweb.org/articles?id=10.1257/mac.20170388
Maybe Houston and Austin are instructive here? Similar climate, similar amenities, similar density,, similar per capita GDP. The salient difference is land-use policies, and the rents in Austin are 50% higher. There's a lot of artificial scarcity baked in to housing prices in some places.
I'm tempted to ask if part of the issue is that the areas with more expensive homes relative to wages are simply areas with more established money.
On some level, in a classic economist "assume a can opener" sense, it's odd to have this difference between areas. If in some areas housing prices were higher relative to wages, you'd expect a movement in wages or in housing prices (or in people moving) to equalize it. Various reasons why that wouldn't happen ... maybe it's reduced costs in other parts of the budget (e.g. cars), maybe it's that they're more attractive places to live so people pay more for a better product, maybe something else.
Or maybe it's that in the areas with higher prices relative to wages, there's an added element of housing being bought with old wealth instead of wages.
In this view, some project city YIMBY utopia built out in the plains in Nebraska would be cheaper than NYC, simply because there's no rich trust fund babies.
I think the issue is that you are treating all types of building as the same, MattY is separating out hones and other stuff. People don't want to live somewhere because of the homes available (otherwise there wouldn't be lots of cheap property in the rust belt) but the other stuff. If you build more homes but keep the supply of everything else low (schools, shops, jobs, parks, etc) the prices will go down. If you build lots of nice new stuff but keep houses fixed price of housing goes up.
Love this step-by-step reasoning.
The big picture concept you're grasping is "spatial equilibrium", the idea that migration equalises quality of life. Quality of life has pull factors like "friends, jobs, museums, and nightlife" (agglomeration) and push factors like rents, travel times, and crowding disamenity. Basically, quality of life equals wages plus fun things minus costs and not-so-fun things, and this residual equalises across space. Imagine water levels equalising when the tide changes.
BTW there's two parts to "equalising". One is the direction: people migrate from worse to better. That's your "perpetual motion machine". Another is the impact: in-migration bids up rents, bids down wages, increases congestion, and worsens crowding. That's what stops the machine.
Example: Oakland gets cheaper and induces in-migration. More people = more bars = more fun, yay! In-migration continues. Bartender wages fall. Rents rise. Traffic gets slower. Bar fights increase. People stop arriving. Oakland is bigger. Maybe it's more expensive, or maybe it's cheaper if the growth was managed badly and had high amenity costs. Falling rents can represent falling demand just as much as they can represent rising supply.
This is spatial equilibrium, and it eats up housing price gains faster than we can induce housing suppliers to deliver them.
We totally need better write-ups of this concept. It's critical for zoning debates, and very underappreciated. Many people assume upzoning will change supply without changing demand. But that's wrong. (Some economists even embed this in models of housing density by assuming a "closed city")
Cameron Murray's stuff covers so many parts of this.
Example: stable rates of rent to disposable income suggest you can't sustain lower rents for long (unless you've ruined amenity at the same time). https://www.fresheconomicthinking.com/p/why-is-the-rent-to-income-ratio-flat
And the parallels between spatial equilibrium in housing and the Downs paradox:
https://www.fresheconomicthinking.com/p/downs-thomson-housing-paradox
And a cracker on AMM models which assume a "closed city":
https://www.fresheconomicthinking.com/p/its-time-to-throw-out-the-standard
Doesn't this theory require induced demand to increase demand at >1:1?
Oakland has 500,000 housing units, plus 10,000 people looking for housing.
Oakland builds 1 housing unit.
Oakland has 500,001 housing units.
What happens next:
1. YIMBY: Oakland has 9,999 people looking for housing, so housing prices fall.
2. Scott: Two people hear that Oakland is a larger city with better night life and try to move to Oakland. There are 10,001 people looking for housing. Housing prices rise.
It seems like Scott's smoothly parsimonious pricing function requires that any increase in housing supply/population generates a >1:1 increase in demand. So if someone moves to an empty patch of land, that should induce two people to want to move there. Otherwise, constructing a second house on that land would cause housing prices to stay flat or fall.
I think you're saying "each marginal unit of housing built increases marginal demand more than it does supply", if I'm reading it correctly. I think this is obviously wrong -- the creation of 10 units of housing does not also create with it 10 people wanting to live there. If we think of it as a ratio of number of units demand created per unit built, the ratio is clearly not 0, but it's also obvious to me (with no evidence whatsoever) that it's less than one. Close to 1, perhaps, but not 1. The ratio also clearly depends on how big the city already is -- it's closer to 1 in NYC, and closer to 0 in the middle of the great plains.
There are also lagging factors here -- if you magic'ed 10 million houses in SF tomorrow, it'll take many years for the population to actually move there (after everybody already there all spread out to their own house). So the game is to build faster than people can move, otherwise house prices will go up as you build because you're not building enough.
Edit: I think your insight of the mobile young'un is broadly correct -- NYC is currently too expensive, so they don't move there. So the demand curve for these cities is actually very flat -- you can build 10% more and houses will barely budge. Another way of saying that house prices are inelastic to demand in big cities.
There is a problem with using median local income in this analysis. Restricting housing construction will generally raise rents which will price out the poorest residents, all things being equal (of course there are considerations like preferences around amenities and floor space and so on). This is part of why SF median income went up so much. Therefore building housing will lower measured median income as poorer residents move in, even though the incumbent residents are as rich as they were before (again ignoring effects on wages and so on). So you might expect housing construction to lower rent (a real benefit) and median income (a fake statistical artifact), thus keeping rent/median income where it was.
A good concrete example would be to think of someone working at Wells Fargo in SF 15 years ago. Even though rent/median income might not be that much higher in SF (growth of tech / low income people pushed out), for the typical incumbent SF resident, rent/income has probably gone up a lot, because banking wages (in this example) are determined nationally. (And then WF responds by moving jobs to NC to keep things in equilibrium). So construction is just this process in reverse - rent/median income will still constant but the typical resident will be better off.
I think what this misses is that demand is both number of people and ability/willingness to pay. The thing that NY has but other places on the eastern seaboard don’t is the global headquarters of the finance world. By connecting into that, the ability to pay goes up. Adding housing to the middle of Idaho doesn’t create a financial center and therefore doesn’t increase demand.
Not that this probably affects the central thesis here much, butfor any discussion of 'demand for housing' it's worth bearing in mind that the cost of housing is to an uncertain but significant degree artifically bouyed by its emergence as an apparently 'evergreen' investment asset. People buy urban property because the value is likely to rise, a prophecy which has become self-fulfilling over time.
I think you’re wrong because you seem to be misunderstanding the causation and some other factors at play.
Housing is not expensive in New York because it’s dense, but because that’s where the money and work to be done are. (Which began with the harbor, the river, etc.) That affects not just housing prices, but salaries, migration patterns, opportunities, culture, politics and regulations, square footage, demographics, etc. And yes, all that feeds on itself to increase demand which puts upward pressure on prices. But only because demand rises faster than supply – which is caused by people seeking money and work, not housing.
It won’t work if you start with the supply side. You can’t build a new New York, or 10x Oakland’s economy, by simply building enough housing.
People don’t move to places with housing, but to places with money and work (even if they can’t get a hold of it, as slums around the world illustrate). Ideally places with work and money would be the same places that have housing, but as so many dying towns, bereft of corner-stone industry, testify to, it’s not. Having the housing doesn’t guarantee the work.
Building more houses where there is high demand for housing *does* contribute to an already hot local economy, but has a net downward effect on housing prices.
PS: Note that new housing is always new, i.e. high quality, built for people with money, often in rapidly gentrifying neighborhoods. Newly built housing is rarely itself affordable housing (to local politicians’ chagrin). But it will make other, older housing, in adjacent or comparable areas relatively more affordable. Even if all prices go up, the prices for the old housing won’t increase as much as they would if there were less housing.
That also means that to understand housing prices for a city, you have to include the cost of housing for everyone who works in the city, even if they commute and don’t live inside city limits. (If public transportation is good, it’s not necessarily a bad thing to live in the suburbs or a nearby town.) We shouldn’t expect the average housing prices in SF proper or Manhattan to reflect the average rent/mortgage paid by the people who work there.
PPS: The difference between New York and Not-New York affects all the numbers: mean, median, modal, top and bottom percentile. I don’t think there’s any good number you can use to compare the cost and value of living in New York, to the cost and value of living on a plain in North Dakota or anywhere like it.
You’d get a better picture if you compare housing prices in different parts of New York State or NY metro area, which can be seen as part of the same market if you squint. And you don’t have to travel very far from Manhattan, before the housing market looks very different.
For good, generalizable understanding, it would probably be even better to use cities and regions that don’t have multiple world-class universities and industry epicenters, that attract people from the entire planet (and so doesn’t have practically infinite housing demand, like NY and SF). Looking at more normal cities will probably give a better picture of how supply and demand works in the housing market.
I was going to say this is a minor technicality, but actually the more I think about it, the more it may be a significant issue. In the early 2010's, the little town of Williston in the plains of North Dakota was actually *the most expensive* housing market in the United States:
https://www.theverge.com/2014/2/19/5425040/williston-north-dakota-most-expensive-place-to-rent-in-us
https://www.apartmentguide.com/blog/williston-nd/
It turns out that oil companies were willing to pay a lot for workers, and anyone who could have worked in construction could get more in oil, so construction got really expensive, and supply was not able to rise to meet demand. They probably developed *some* amenities that weren't there before, but the rise in price was really supply and demand, rather than the agglomeration effect for amenities.
The rest of Japan doesn't need good housing policy - it has declining population. Japan as a whole has declining population, but there's still some people that haven't moved to Tokyo, and as they do, Tokyo is still slightly growing (I believe).
Obviously, the time scale matters here. Certainly, if you simply disallowed anything more than single family dwellings you could prevent a sleepy town from becoming a major metropolis. So, certainly in the long term, there can be such a thing as induced demand (though whether this holds in expectation is unclear...not all possible cities will become cities). And what you are doing there is basically making the city less of an attractive place to live than it would have otherwise (maybe not to the people who want to live in a sleepy village but to the average person).
But, in the near term, we can model the amenities/jobs in the city as relatively constant. Or, to put the point differently, no one is deciding where to move by looking at the total number of units in that metro area. They look at the amenities and the price.
Thus, the only plausible way you draw more people to the city by building units is if house prices go down. In other words, absent the price signal what induces demand?
Now, in the longer term, more people draw more buisnesses, more restaurants etc etc... So certainly in the longer term you can induce demand but that happens after more people move in so on a longer time scale.
So yes, it's plausible that in the long term more housing leads to higher prices but only via reducing the prices on the margin in the short term.
Edward Glaeser (Triumph of the City) here would point out the relevance of the Z-axis (building height) and regulations preserving building stock, and in particular low building stock. San Francisco doesn’t have a lot of tall buildings. It has a lot of historic low buildings. So too does Manhattan, which brought us the elevator but also brought some of the preservation orientation that you are probably seeing in the data. Both cities could be denser and taller than they are. Enrico Moretti at Berkeley (where I teach) has a good paper on this that I will find and post a link to.
Japan has experienced a demographic transition, with an older population and few children, leading to negative population growth. There are whole villages where housing is very cheap since their owners have aged and their children are uninterested in the rural lifestyle. It's interesting that the problem is always "not enough houses (or water, or food, or jobs)" as opposed to "too many people".
Noah Smith did a post addressing this issue (or something very like it) about a week ago. (https://www.noahpinion.blog/p/luxury-construction-causes-high-rents).
Bottom line: more construction doesn't cause higher prices, ever, except in the sense that demolishing a 20-unit "average" apartment building and replacing it with a 20-unit "premium" apartment building will cause rents for that building to go up. There will also be a small effect on the immediate neighborhood, causing rents in nearby buildings to increase slightly due to the neighborhood being seen as more "premium". But housing prices in nearby neighborhoods will go down, because there is more supply of housing in the city.
You're wrong t think of housing prices as being driven primarily by housing density. Local desirability is far more important. In my part of Connecticut, house prices in low-density Avon are much higher than high-density Hartford or New Haven, because people with lots of money to spend on housing want to move to "desirable" communities - and one of the prime factors in making a community desirable is the quality of public schools. And public school "quality" (as measured by things like graduation rates, SAT scores, absenteeism, etc.) is influenced much more by the students attending the school than by any action the school teachers or administrators can take. So, house prices are high in rich towns because rich people want to live in the rich towns, so their children will associate with the children of other rich people, and absorb bourgeois expectations.
Re-consider your Oakland thought experiment: If Oakland were to build 10 times as much housing, why exactly would people with high housing budgets want to move in? They'd be moving into a hard-luck seaport city with poor schools and high crime. It wouldn't be for the awesome urban amenities, because Oakland doesn't have them. And no one will build them until the yuppies (or whoever has the money to demand the amenities) move in. Which won't happen because there's nothing to draw them in.
Extend your thought experiment to that wide-open plain in North Dakota: If you build a dense city with 3 million housing units to support a population of 5 million people, who exactly do you think will move in? Aside from the oil field workers. Would San Francisco tech workers move there as a more affordable alternative to San Francisco? If they want San Francisco's amenities, they won't move until your instant North Dakota city has the amenities. And who will build the bars and coffee houses the tech workers want if there are no tech workers for customers?
I think you have to separate out *housing* density from job and amenity density. The places you're describing have a lot of both. But they're not the same thing. Think even more micro-level and this is obvious. The most expensive real estate in any given area tends to be in a central business district with lots of jobs and amenities but not an especially large residential population.
If you increase housing without adding more jobs or more bars or more museums, then you're not inducing any more housing demand. You're just creating supply and prices will go down. There's a tendency of course for housing to induce amenities which induce housing demand and so on. But if you just build a bunch more housing (especially in an area that's already pretty saturated and doesn't have space for new amenity development), you're not setting up that circular dynamic. A lot of the big ticket amenities are fixed in the medium-run anyway. You can double the size of NYC, and they're still not going to build a second Met or bring in two more baseball teams.
There are some weird, off-diagonal places out there to confirm this. City of Industry, California has a population of 264 and a density of 17/square mile, which sounds rural. But it also has 67,000 jobs, and it's quite expensive there. On the other hand, the densest cities in the US technically speaking are several New Jersey suburbs of NYC, all of which are much cheaper than the city itself because they don't have much to draw people in. The slums of major cities in the developing world are some of the most densely populated places you'll find, but they've got amenities and they're very cheap (consider that the most densely large populated cities in the world are Manila, Dhaka, and Kathmandu).
If Oakland builds only so many new houses that all the related factors of attractiveness would be perceived the same (diversity and density of jobs , higher quantity and quality of cultural/ social/ medical infrastructure etc.) I believe *this one* specific activity would lead prices to go down or at least to stagnate. The problem might be, that if Oakland builds only so many houses & the level of attractiveness stays exactly the same, the demand-effect of people wanting to move to Oakland is already surpassing the effect of the additional houses. So in effect prices go up. However, they do not go up as much as they would have without the additional homes. So Oakland, even if they were the only ones to build new homes, in this scenario is better off than if they didn't. (Except if they are perceived to be the only place around with a great YIMBY culture and sensible politicians, which might also increase attractiveness. But then it's *not* purely the number of homes again.)
I believe there are thresholds, and the attractiveness of our exemplary place O. rises only if there are so many more people, that this attracts, say, a new university, two major industries, a new high quality theatre ... or at least lots of new restaurants and cafes and local arts culture (I don't know Oakland, so using examples from my area). In other words, if it is perceived to be a bigger *and* therefore more high-quality / more opportunities / hipper/ ... place.
I would be interested to see good data, how density correlates with housing prices in my country. And how city size correlates with the latter. Or what about European cities ... okay, that's maybe a bit unfair.
And yes, I think building new homes makes sense, but I think housing prices need to be tackled by other means.
>Or doesn’t it (as Yglesias argues) allow an economic perpetual motion machine, where you just keep building houses and generate infinity money as the price of each keeps going up?
Apropos of nothing, the canonical Urban Land Rent model in Econ, Alonso-Muth-Mills, assumes all workers labor at the platonic center of the city, an infinitely tall skyscraper which uses no land. I often think of this building and its employees.
Here's why you're wrong in a single sentence: Demand causes high prices, not new units.
Prices are high in SF and NYC because those are desirable places to live for a huge number of people. People all over the country and the world would live there if they could, and prices reflect that. The fact that the densest cities are the most expensive is true. But the high prices are not caused by density - rather, the density and the high prices are both a consequence of crushingly high demand.
I generally agree with your hypothesis. More housing is more supply which should reduce prices. But a simplistic Georgist analysis would seem say that more housing increases the value of all the other property around it, which includes other housing. It doesn't seem impossible that in certain circumstances, the 2nd effect could outweigh the 1st.
I'm curious if any Georgists would like to weigh in on this one?
You vastly overestimate the percentage of Americans that want to live in a dense city. Some Americans do, but most don’t, so density is at best a wash and probably reduces demand. Thus, the causation flows almost entirely in the other direction, with high demand causing high prices which cause density.
When considering the price of a thing, demand is an important variable that can not be reduced to supply alone.
There are volumes of ink that can and have been spilled about the details, but you're not getting this to reduce to a function of one variable.
The argument is that, on the margin, increasing supply will reduce price of a particular good. An observation that the price of that good is currently highest where the total supply is highest doesn't really engage with the argument. Second-order effects are table stakes.
What about the following theory, which is consistent with both your arguments and the studies: if you build houses, prices will temporarily decrease, but then they will increase later after some amount of time (e.g. 20 years).
The delay in the price increase comes from the long time it takes for density effects to materialize: people don't just want to move to a dense city, they want to move to a city with jobs. However, it takes employers time to set up in a new city, and therefore, the increase in employment opportunities that follows from density might take time.
(As an aside, if people want to move to dense cities due to jobs and if those jobs materialize disproportionally in dense cities, then it is a bit of an economic perpetual motion machine; the efficiency gains from density must be *huge* to justify such a story.)
I found this a bit upside down and tried to put down some notes. Not ready yet, but as some points seem to recur in discussion ....
First, what if what if density doesn't drive prices, and prices do not drive density? What if instead, there is a third underlying variable that affects both. That variable might be attractiveness. Okay, the text mentions this.
Second, but what concretely is making a place or city attractive? Elements might be: the status of the city as (local) capital, the natural environment in the surroundings, the city's image as being hip or you name it, the amount of available jobs and career prospects, the quality of cultural infrastructure (theatre or whatever you are into), the perceived likelihood to find like-minded people, etc.
Third, what then is the relationship between attractiveness and density? I believe beyond a certain threshold density as such is not attractive to (most) people. I believe desity is an unperfect proxy for some of those factors of attractiveness (availability of high quality theaters) - and unrelated to others (beauty of the natural surroundings).
Forth, instead of calling it a proxy, one could also describe a feedback loop that involves density somewhere: more people moving to a place leads to higher density, this leads to more industry being attracted and more needs for culture, which leads to more jobs & more new theatres set up, which leads to the place being more attractive, ...
Fifth: why is this important: Because it impacts the hypothesis that more housing leads to more density leads to higher prices. Whenever more density doesn't lead to 'higher *perceived* level of attrativeness', then in this model, it also shouldn't increase prices.
Sixth: But what about the feedback loop? Well, it is probably happening in many areas, but there need to be other factors involved, and it doesn't need to happen always. Among others see next:
Seventh: There are probably also thresholds. Not every 1% of new housing leads to more attractiveness factors or the city being perceived as 'bigger'.
Eigth: Isn't city size the factor and not density? I suppose only few people really want to live in a dense place as such. Maybe city size is a better predictor for those attractiveness factors? Would be interested to see if prices are correlated better or worse.
Ninth: Correlation is not causation. That's clear to everybody here I suppose, but I had been missing a more explicit follow-through on this in the text.
Tenth: Why *perceived* level of attractiveness? It's all in the eye of the beholder.
TL;DR More housing lowers rents immediately. More density slowly increases rents.
I think it's important to be a bit more careful about the timing/delay of impact when thinking about the different steps in the causal diagram here.
Here's my mental model:
A) More newly built housing -> People in already in city consider moving to that newly built housing
B) Rent prices come down (on quality adjusted basis) -> People in other cities consider moving here
C) More people living in the city -> demand for the city goes up, since some people like living in denser places with more people and more businesses
D) Many high income people -> new businesses open that cater to high income preferences
E) Many high skill people living in a place -> more high paying employers set up offices in city
F) Many high paying jobs -> people want to live in the city (housing demand goes up)
In terms of how quickly these channels operate. I think A>B>C>D>E>F
Channel A is the fastest to act. So new housing very immediately affects the set of options of current residents and starts to lower rental prices (relative to what they would have been under the status quo).
This is partially offset by B, with people seeing the lower prices and considering moving to the city.
Channel B can at most partially offset the impact of A (since by definition B is people who are moving *because* the rental prices are lower). Considering moving to a new city is a slower process than moving within a city and as a result, this channel also generally reacts more slowly than A.
Channels C + D are related to the impact of the changing population on the desirability of the city (either directly through the influence of peers or via new businesses that open).
Both of these are even slower processes. It takes time for the population to expand in response to the new housing and for people inside/outside of the city to notice the changing demographics. Setting up a new business is also a bigger investment that takes more time compared to a family moving between apartments. Also notice that these channels only make sense if the population is growing, but if B is growth caused by lower prices, then it's still hard to see how this could lead to a net increase in rental prices (relative to status quo).
Channel E (employers choosing where to locate offices) is much much slower. Many cities wish they could convince more high paying employers to move to their city, but it takes a lot to overcome the inertia. Also high rents for housing and office space directly push against the desirability of a city to employers, since those add to the cost of moving to the city.
Channel F (people moving to areas with good jobs) doesn't happen until E happens, so it is yet more delay in the response.
In my view, the new housing lowers prices* in less than a year after the new housing. All of the other factors kick in much more slowly over a period of years, and at most partially offset the decrease in rents.
* on a quality adjusted basis, relative to the status quo of not building.
Channel D
Not understanding why you are only talking about reverse causality as a possible challenge to your interpretation. What about omitted variable bias: there is some other underlying factor that is associated with density, which is in fact driving higher housing prices. I would tend to think that OMB is a huge problem here, as there are many very dense slums in the world that are not in fact not expensive at all, most famously Kowloon City.
I have a few guesses as to the origin of OMB, most importantly, jobs. There are certain locations that are just much better in terms of jobs. If you want a job in the media or finance, you should probably move to New York. If you want a job in government, you're going to have to move to DC. If you want a job in the movie business, LA is really your only choice.
If jobs are in fact driving much of the increase in housing demand, building extra housing is not going to raise housing prices. Moreover, if these are relatively high paying jobs, then people are still going to come even if you don't build anything, because they'll be able to afford more than the average resident in the area. And this is exactly what has happened in the bay area: the working and middle class has steadily been pushed out by tech professionals.
This seems like a far more natural interpretation than people move to an area simply because there are a ton of other people already in that location. New York's density is a strike against it in my book: too noisy, too crowded, too stinky.
My guess is that what you're really thinking about is agglomeration effects, but returns to agglomeration come from industrial agglomeration, not from residential agglomeration. It's a critical mass of businesses in a given location that makes it into a jobs hub and attracts migrants.
I think this this analysis suffers from three problems:
(1) Reducing a complex multiple variable and step interaction to two factors (housing density & cost). There are a large number of motivation factors which you touch on but then discard from your model to use density as a proxy(Jobs and income being the most significant).
(2) confusing a correlative measure (density) for a cause. Density correlates with cost because they are both result from the same cause (Jobs and income). If stripped of its jobs, Manhattan would have low housing cost, despite the high housing density. Conversely, If you built 5 million units in Oakland or the North Dakota, the cost plummet, at least initially. This transient phenomenon only makes sense if you include the omitted variables.
(3) Explaining examples of urban population decline. There are many cities which have experienced population decline due to loss of industry. If population density were the primary motivational driver, this would not be the case.
At best, it can be said that people follow jobs, Jobs & employers follow population density, and higher housing density and housing cost both *result* from competition for increasingly concentrated locations.
If jobs opportunities are held equal, people will flow from high cost to low, not the other way around.
I am skeptical of the two charts you showed at the top because: (1) They use survey data when there's better data available. (2) The category "Urban Area Corresponding to Major Metro" seems like it allows the authors some choice in what to include or not include. The second one also seems a bit arbitrary - including San Diego & Cleveland, but not Minneapolis. (3) Neither of these includes "Podunk" or "Empty Space", so connecting them to the hand-drawn chart is extrapolation.
Is it better to use housing price or housing price divided by income as the vertical axis? I could see arguments in both directions. Both housing price and income probably change as more housing is built.
I am also confused by the relative positions of Miami and San Francisco in the two charts. Was Miami significantly more expensive than San Francisco in 2015 and significantly less expensive in 2019? If no, then the error bars on the data are large enough for me to not trust the trendline. If yes, then maybe someone should figure out how to do what Miami did.
I'm considering remaking these graphs using Federal Reserve Economic Data: https://fred.stlouisfed.org/release/tables?eid=1138280&rid=462 . Let me know if you think this would be valuable.
Some initial observations from looking for the highest things on the list: (1) The most expensive core-based statistical area is Vineyard Haven, MA, a little island off the east coast. Manhattan might be even more expensive - it's not listed separately - but Martha's Vineyard is more than twice as expensive as metro New York City. (2) San Jose, San Francisco, Los Angeles, and San Diego are all in the top 25, but so are Santa Maria, Santa Cruz, Napa, Salinas, San Luis Obispo, and Santa Rosa. It seems like California (or at least coastal CA) is unusually expensive, even for small cities. (3) Hawaii and Rocky Mountain ski towns (Jackson, WY, Breckenridge, CO, etc) are also unusually expensive.
My guess is that there is still somewhat of a positive trend between housing price and city size (or density), but that it is noisier than these graphs show, especially for smaller cities. It should be possible for a city which is currently surprisingly expensive relative to its size to shift down and to the right by building new housing. For example, on the top graph, if San Diego became like Las Vegas.
I agree that Tokyo should not be placed on these graphs. But it would be interesting to see where Tokyo lies on a graph with other Japanese cities. In a similar graph for Japan, is the slope positive or negative?
I can't follow your logic. Primarily cities are composed of a few moving parts. Primarily there is a source of a harbor or natural resource that fuels primary demand. Think of the ports in Chicago (and the stockyards during the turn of the century), or other major port cities. Indianapolis' main driver of economic growth is its interconnection of highways for example. These are elements of any city that can't be changed or relocated. The Chicago mercantile exchange is in Chicago because Chicago outbid St. Louis for stockyards (fascinatingly because they put in a sewer system) and because they have access to a river and ports. Surprisingly, not many people know that Cleveland, Ohio used to be much bigger than Chicago. This maps onto countries as well. Countries that are farther away from ports tend to be poorer, while countries that have large amounts of natural resources and are historically non-European tend to be resource cursed to exploitation.
If it were cheaper and more efficient (ie made more money) to have the Chicago Mercantile Exchange in Pidgeonforge TN it would be in Pidgeonforge TN. But it isn't.
So, cities that have a primary resource that can't be relocated tend to have a primary driver of economic growth that is stable through economic fluctuations. Software development can done anywhere, and often is, but the port structure of a Seattle or San Francisco won't change.
As to the cost driver of housing primarily? House prices in the long term appreciate in those places where costs are expected to go up. And prices are more likely to go up in those cities, over a ten, twenty or thirty year term, that are near resources that are structurally economically self sufficient. Were there some way in which Topeka Kansas had a structural comparative advantage over a port city on the coast then people would all move to Topeka Kansas and crowd around that source of economic development.
The farther you drive away from a major port the less expensive housing is, typically. Meanwhile ports tend to be geographically landlocked or have limiting geography, such as the LA basin, New York, or San Francisco.
Port cities are comparatively expensive because the ports are continual sustainable engines of economic growth that don't migrate with people with the fluctuations of the economy. That's it. There's not a whole lot of mystery here. This prisoner's dilemma unstable state Nash equilibrium notch is largely incorrect.
The NIMBY problem makes housing prices worse in those areas where people take over zoning commissions, but in cities that have a port housing in the long term will remain expensive. In cities without a port NIMBYism is much like price controls - all it will do is disincentivize housing even further and lead to urban blight. The less affect a port (and it's subsequent markets) has on sustainable long term effect on urban growth the worse NIMBYism will be for the economy because during a downturn prices will remain high. This is why NIMBYism in San Francisco is comparatively worse than say, NIMBYism in New York or Chicago as its economy is more heavily dependent on the tech sector which is highly cyclical. High prices during a downturn lead to people moving away, lead to empty store fronts and then urban blight which becomes self reinforcing.
You're overthinking it.
EDIT - On further review, I can't tell if you're joking or not. You could make everyone richer by burning money and causing a small amount of deflation, but I wouldn't recommend it. This looks like idiot bait.
I think what this argument is trying to show is, if Oakland got 10 times bigger, *with the same restrictions on housing development*, then the price of housing would go up. Which I think is correct, but it's controlling for the wrong thing, because loosening restrictions on housing development is precisely the mechanism that YIMBYs want to get more housing from. (One question about the scenario in the post is why people would build more housing if it's not easier to do so.)
What YIMBYs are saying is that reducing restrictions on housing leads to more housing and lower prices. This is clear from thinking about the supply and demand curves. In the absence of artificial restrictions on housing development, we would have a nice supply curve that goes from bottom left to top right. With restrictions, developers can't build as much as they would like, and basically the curve hits a wall at some amount of housing units. So the equilibrium with restrictions occurs at a lower quantity and higher price than the equilibrium without the restrictions.
The 1st order effect of loosening the restrictions is to move the situation closer to the equilibrium without the restrictions. If you read "The Economic Implications of Housing Supply" by Glaeser and Gyurko (2018), you see that the housing cost doesn't go too far above the minimum cost to build, if your restrictions aren't too bad and the housing market can adjust to demand (their example is Atlanta). For the Bay Area, the price is more than double the minimum production cost (these are both shown in their Figure 2). So there's a lot of room for prices to go down in places like the Bay Area.
What you're pointing out here is that there's a 2nd order effect, where having a bigger and denser city makes the demand go up. This is true, but I suspect that the effect is just smaller. "Growth, innovation, scaling, and the pace of life in cities" by Bettencourt et al. (2007) shows that if you double the size of your city, a lot of metrics around socioeconomic efficiency of the city goes up by ~15%. So things like your GDP per capita (and crime rates) grow by ~15% if you double the population, which explains why people want to move to big cities, and would make it more expensive to build housing (higher labor cost) and make people willing to pay more for housing, but it's not big enough to counteract the 1st order effect.
There's probably better studies to support what I'm saying, but these are just things I can think of off the top of my head.
You're working with a theory that there is a pool of people who want to live in big cities, but individual big cities are to some extent substitutes for each other. This pool can be overwhelmed with sufficient supply, but if it is not overwhelmed then big cities will be expensive *because* they are big - there is unsatisfied demand for dense living.
I disagree. I think people want jobs, and they move to big cities because this is where the jobs are.
"But that's just adding another step! The jobs are in the big cities because that's where the other people are!"
To an extent this is true. You can set up a sandwich shop in Sydney and sell a lot more sandwiches than if you set up a sandwich shop in the Great Sandy Desert.
BUT it isn't enough! Each additional person creates LESS THAN one extra market-rate job. If this was not the case and density fed on itself in an ever-increasing way, we would see cities grow ever bigger and bigger, with the goliaths outcompeting the giants. Why do some people live in SF rather than New York City? It's not because SF is a discounted option relative to the more populous and therefore more desirable NYC.
As a further counterpoint, consider declining cities like Detroit. If density feeds on itself, and it is the mere fact of more people creating more demand, then these cities should have not depopulated. Detroit was bigger than many other cities that have not shrunk so it's not simply a scaled-up example of the broader urban immigration phenomenon.
What you're missing is that a city needs an underlying economic engine. This is very obvious in small towns where there is often one dominant local industry, and the whole town's fortunes rise and fall with it. Not everyone in Nangwarry works in the forestry industry, but it's the health of the forestry industry that decides what housing demand is like and how many sandwiches you can sell there.
This continues to be true even for mega cities like NYC or Tokyo. Yes, they have so many people that there are lots of jobs supporting all the other people there, doing things like cleaning, car repairs, etc. Yes, there are so many people that they are good places to set up industries that could be based anywhere but want a network of similarly-skilled people in close vicinity. But there is a fundamental driver of why these places are big - they have a geographically fixed multiplier of economic value that is extremely powerful. That engine is a port.
There is a reason that all of the world's biggest cities are located on coastlines, and it's the reason that Adam Smith identified in The Wealth of Nations. Sea trade is extremely efficient. If you have a port, you can cheaply trade with every other city that has a port. This in turn allows you to unlock greater division of labour and greater efficiency than would otherwise be possible.
That's not the whole story of course - you need to have things to trade through the port. But it's a key reason why everyone in SF and NY doesn't just move to somewhere like Austin where they're cool with building stuff and they can have a super-dense mega city.
I think D.'s and Robert Donnelly's comments together explain it well.
You're missing an intuitive understanding of causality that maps back to a logical framework. I recommend Judea Pearl's Structural Causal Models, which he explains simply in his Book of Why. Note that the Book of Why is imprecise in a few parts, but it's a good enough start. The "correct" explanation is given in "The Do-Calculus Revisited," also by Judea Pearl, though it's a harder read.
I want to address 2 points, the long term correlations vs short term studies and the potential gains for a defector city in scotts hypothetical.
For long term correlations, recall that its possible to be a failed city. Baltimore was the 2nd largest US city in 1840, now its 30, and not just because a bunch of western cities overtook it, many eastern ones did to. In regards to induced demand, if a city is a long term success, it will have many houses and high prices ie NYC. Within some margin of acceptable housing policy it could be expensive relative to its success or cheap, but still bound to that curve. The studies show how a city can overperfom its position on the curve. But it would be foolish to think you could easily control a citys overall success. A nimby policy on the premise of capping housing prices by capping city success is likely to become baltimore cheap, not upper west side 30
years ago cheap.
Success is relative. The defecting city would see short term increase in housing prices as it under perfoms its spot on the curve. Long term will be driven by success but intentionally restrictive policy is likely to cause the bad kind of price drop.
Suburbs dynamic is different because its success is largely a function of the nearby city thus exogenous. It can have no housing growth but still be very successful and therefore expensive.
I see statewide control a la CA as not primarily resolving a prisoners dilemna between big cities, but forcing suburbs to upzone along with the successful city they are attached to.
Finally, success is also good. The YIMBY appeal isnt just globally lower prices, you get 1) more successful city 2) overperform your spot on the curve. Win win.
how do we account for slums? clearly it's possible to build housing at very low cost in very high density locations, in the absence of any regulations.
If you believe that, at the margin, building a thousand new units of housing in Oakland would cause housing prices to rise because of the induced demand, then you must logically believe that at the margin, tearing down a thousand units of housing would cause housing prices to *decrease*. Do you?
I'll even throw in the explanation of why it "should": some of the people living in the houses that were just torn down will say "screw this, I'm moving back to Stockton", and some of the people on the outside planning to move to Oakland will say "meh, the city doesn't seem so hip and cool now that they tore down those apartments, so now I'm rethinking this move". Those are things that will absolutely happen. They are the mirror image of the things that will happen if new housing is built, that lead you to believe that building new housing locally increases rents.
I'm pretty sure you recognize that this is false - there will be *some* people who abandon their plans to live in Oakland if you tear down a thousand apartments, but there pretty obviously won't be so many of them as to drive down rents at the margin. And at the margin, the difference between -1000 units and 0 units, is going to be very similar to the difference between 0 units and +1000 units.
Alternately, I look forward to your argument that Oakland should help its fiscally stressed renters and can't-quite-afford-an-apartment homeless by tearing down homes.
I think the essential chain of causation is something like:
1. Faster economic growth causes higher prices
2. Higher prices cause higher density (and other negative externalities)
3. Higher density (et al.) causes slower economic growth
4. Slower economic growth causes lower prices
So you have a dynamic equilibrium. Density (and other negative externalities) is why economic growth and prices in clusters like NYC or Silicon Valley don't just explode upward endlessly. Density is a brake on growth and prices, not an accelerator. It is correlated with higher prices because higher density is needed to fully brake the growth of a region with greater economic potential, but the density doesn't cause higher prices but the opposite
> Wouldn’t the alternative be some kind of highly unparsimonious pricing function like this?
I don't get the argument of this graph. Doesn't your imaginary graph look like you're talking e.g. about Minneapolis-St Paul (on the real graph) and just pretend that all other cities fall exactly on the line?
> This is a coordination problem: if every city upzones together, they can all get lower house prices, but each city can minimize its own prices by refusing to cooperate and hoping everyone else does the hard work.
Well, you can also spin the opposite narrative just as easily!
Assume cities like high land prices (eg because it gives them more tax revenue, if they use property taxes or land value taxes). Then each individual city can get more revenue by building more, but overall all of them together might perhaps not make more money.
Building more would be a 'tragedy of commons' in this situation.
---
Overall, I think your analysis is more-or-less correct, but that property taxes (or even better land value taxes) can fix the incentives for cities, so they can build up without having to coordinate with other cities.
Yes and no... the book I would point you to is Alain Bertaud's "Order Without Design" if you want a good, reasonably accessible foundation in how the economics of cities work.
Essentially a city is like a big party - it's a huge market that becomes exponentially more productive the larger it grows... so long as the city can effectively mitigate the downsides of proximity and density and avoid having the party fall apart after a few crucial guests leave. The biggest issue is maintaining mobility - most US cities are car-dependent. Car dependency allows cities to sprawl out and provide a lot of living space per person, but it also makes moving around very difficult above certain densities - thus making it hard to build an effective city of more than a few million. To get bigger, cities have to get more creative with their transportation networks.
New York and San Francisco are big, very successful parties that everyone is trying to get into. In one sense, they are unaffordable because it's very hard to get into them and allowing more people in would make them more affordable. But in another sense, it is the fact that so many productive and attractive people are clustered together in these cities that makes them so attractive and hence unaffordable.
It is certainly not true that building more housing makes prices go up. But to strong man this argument: ultra high prices are a drag on growth in very productive and attractive cities. Building more housing will lower prices, which will in turn allow more people to live in the city, further increasing its productivity and in turn attracting even more people, causing prices to eventually go up. Sort of like if you start turning away guests at your party just because you don't want to have a big party. Turning away people doesn't make your party more accessible (it makes it more exclusive obviously), but maybe it will stop the party from really gearing up and becoming even more popular.
So the real conflict is over urbanism: do you think it's a good thing for lots of people to be able to live together in cities and benefit from them? As a city planner and urbanist, my answer is yes: I think cities provide all kinds of benefits and I defer to people's expressed desire to live in cities. Outside of Manhattan, I would argue most US cities aren't very dense at all and that there's no reason to be afraid of letting them densify and shift towards different forms of transportation. Right now I think we have a chronic undersupply of the kinds of urban places people want to live in and that are capable of providing satisfying and high-paying job opportunities and life experiences in an advanced economy, so I support letting people build more cities.
Noah Smith had an excellent article last week covering just this topic:
https://www.noahpinion.blog/p/luxury-construction-causes-high-rents
He cites plenty of econ papers, going so far as to explain why the ones that concluded the opposite were mistaken or what they failed to include. I think you can start with his post and its links, and go as deep as you like to satisfy yourself about how it all works.
The key is that no city is a closed system: people move to different neighborhoods within a metro area, and new arrivals to the metro area have a choice as to which neighborhood they go to. Building new housing in one of them causes average rents to go up there because (A) you just built new housing, which is probably priced above median itself, and (B) the signaling that this is a good yuppie neighborhood causes follow-on moving. However, that construction and those moves *prevent* gentrification elsewhere in the metro area, because the people moving to the desirable city are able to buy new-construction housing rather than apartments in an older locals neighborhood. And the yuppies don't *really* want a 100-year-old apartment anyway (they'd rather the new construction), they'll just take it if that's all that's available in the neighborhood they want to live in, and they can afford to outbid others.
Are new builds like old builds? What's the elasticity of demand?
For example, in Pittsburgh's recent revitalization many of the new apartments are swankier and more upscale than lots of the old construction. A good deal of the renovation notably increases amenities. So yes there's more housing capacity but all the new capacity is more expensive for real reasons. So if you build new housing that's more expensive than average and people move in, you can raise the local price even as the movement of people decreases demand elsewhere. (And in fact, this is exactly what you see in Pittsburgh. Sure the new stuff is cheaper than LA/SF/NYC emigrants are used to, but now all the other prices have room and pressure to grow.)
If you build housing that resembles an average or a typical spread, you'd imagine it would decrease or at least not increase the price. You could make an elasticity argument a la Malthus, and argue that the price will fall, then some new marginal people might move in and the price would stabilized where it was previously because there was some reason it was that price in the first place.
To expect the local price to go down purely on the weight of new builds, you'd have to build so much housing that you overwhelm the elasticity of demand. In a place like Pittsburgh whose big appeal is "a cheaper silicon valley", there's probably significant elasticity! The cheaper you make it, the better that appeal! That could be an unrealistic amount of housing! If your goal is to lower prices you're better off making a structural change like WFH or public transit.
I feel like there should be a solution here where we let new development pay off renters in the area. For instance, maybe we give renters in any zone that approves new building in a year a share in future increases in overall property tax. It probably wouldn't have to be a very large amount before it convinced them it was worth allowing building (even if they may have to move away they keep that income stream).
Alternative hypothesis: New York City and San Francisco (and maybe also LA) are weird exceptions because they control the commanding heights of the culture, and therefore are viewed as uniquely desirable places to live. They are desirable not because they are high density, but because they control the News Media / Social Media / Entertainment Industry respectively, and thus occupy a disproportionate amount of mindspace, making a lot of people willing to overpay to live in them.
Chicago / Houston etc don't occupy this kind of privileged position in the culture, and see much lower housing prices.
How does building attracts more people to the place if the marginal effect is raising housing costs?
Like, if your city has a market settled at 100 housings going for $100, that means at that price there's only 100 people that want housing there. If there were more than 100 people willing to pay $100, the price would go up until the market cleared.
If you build 1 housing and that makes the price go up to $101, now there's less people that want to live there than before! Building houses would make cities shrink in population.
The demand curve always slopes up.
If you built 10 houses on a North Dakotan plain and no one bought them, wouldn’t the price go down?….until such time that at least 1 or 2 people bought in…at which point the smallest sight of demand might at least plateau prices at a new (but below initial) level?
It seems to depend on (at least) an ill-defined balance of land availability vs scarcity, location desirability, and hence the result supply vs demand, to even begin to speculate on an answer. My suspicion is the relationship will not be linear. Adding 8 million people in Oakland doesn’t automatically bring with it Manhattan prices, unless you have with it manhattan amenities, and Manhattan geography.
Hold on; if this argument holds it is _exactly_ an argument against YIMBYism: YIMBYs harm themselves and those around them. (In housing prices. Maybe the better quality is worth it.)
Yes, it's better if everyone agrees to enforce YIMBYism globally in this model, but that's not how it works, and it does mean that people who oppose YIMBYism in their own location are acting economically rationally.
seems a silly argument that density is the relevant variable. Demand is multifactorial and is the relevant metric.
I think what's missing here is the type of housing being built: is it single family or apartment AND is that dense apartment market housing? Non-market housing isn't affected by the market as much and can be cheaper. It's also not profitable so isn't common.
If building houses raised prices, the upper-right quadrant of these two graphs wouldn't be empty:
https://twitter.com/JeremiahDJohns/status/1616082065455468547/photo/1
A possible counterexample here is Seattle. Like SF, Seattle's seen a massive amount of immigration over the last few decades, to a large extent of high-earning tech workers (in fact, proportionally more people have moved to Seattle than moved to SF in the last 30 years, both to the city itself & to the metro area). However, unlike SF, Seattle & the surrounding area also allowed a ton of construction, nearly all of which involved/involves large density increases on already-occupied land. The near-complete reconstruction of South Lake Union is the most striking instance, but you can see this happening all over the city (f.e. the ongoing transformations of Roosevelt & Northgate resulting from the opening of the most recent Link extension). Notably, Seattle's median rent also hasn't risen nearly as much; said median rent is far closer to the national average (possibly even slightly below at the moment, if Zillow's rent numbers are to be believed).
One important thing to note here is that Seattle's recent increases in housing supply have been mostly driven by large local increases in density. In some ways, the area faces the same kinds of geographical problems SF does (if not to the same extent), being quite hilly & hemmed in by water. In this, it's unlike the sunbelt cities that frequently get cited as evidence for "building housing to accommodate immigration keeps housing costs down", since there's little in the way of land to build on that doesn't already have something on it.
Possibly contributing is that the upzoning in the area hasn't been limited to one municipality. Several of Seattle's suburbs have also gotten in on the act, with cities like Redmond & Bellevue also seeing enough apartment-type housing construction to radically reshape (or, in the former case, practically create from whole cloth) their downtown cores. To the point about this being a coordination problem, perhaps part of Seattle's success is that the city & its suburbs managed to at least somewhat coordinate here.
Think of China's ghost cities / apartment blocks. Prices surely can't be that high there. Maybe the answer is that developers are good at their job, and build supply where theres demand for it?
A possibly helpful analogy:
English is the language with the most worldwide speakers. English is also the language with the most second-language teachers.
Does that mean that adding more teachers is an important way to increase the attractiveness of a second language?
Maybe, in extreme cases, but not really.
Convenient teaching is helpful, but not decisive, for people choosing their second language. Plentiful city-center housing is helpful, but not decisive, for people choosing their metro area.
In both cases the teaching/housing is a lagging variable that responds to, but only at less than 1-for-1 contributes to, the underlying desirability.
(BTW, Scott, I keep feeling that I haven't quite grasped your actual uncertainty, and I'm wasting both our time on things you already get. That's frustrating for me, and I bet for you too.
Maybe you can restate your thoughts more precisely in a week, after you've read everyone's responses?)
The problem is price and quantity are jointly determined by demand and supply. The exercise that Yglesias is proposing is akin to saying “imagine supply increases then price must decrease to a new equilibrium where more quantity (housing) is traded.”
This is nothing more than the standard Econ101 law of demand applied to housing markets.
The correlations that you show are a big no-no in econometrics. You don’t just regress quantity on prices (or viceversa) and give that anything close to a causal interpretation. Why? Because demand and supply jointly determine prices and quantities. Large traded quantities might be associated with high prices if demand is high relative to supply, low prices if supply is high relative to demand, or anything in between.
The most reasonable explanation for the correlation we see is that places in where supply is large demand extra large (NYC), so prices are high. In contrast, in places with low supply of housing demand is non-existent (North Dakota countryside), and therefore prices are be low. If you run a regression of prices on quantities you’ll get a positive coefficient, but the effect has nothing to do with quantities and prices and everything to do with the fact that in places where supply is high demand is even higher, and in places with low supply demand is close to zero.
If we go meta and ask why is demand so much larger than supply in NYC, this argument naturally ties to YIMBYism. The reason is that supply is artificially constrained in NYC through red tape and regulation. Removing this red tape and regulation will increase supply and drive prices down.
The key fact your missing is if you wave a magic wand and 10x San Francisco you wouldn't 10x all jobs. You would 10x the # of waiters, and garbage men but you wouldn't 10x the # of 500k/yr Google site reliability engineers. And it's the latter not the former that are driving up prices.
I mean, yes, by basic rules of supply and demand. But only in our world.
Most of America is very, very sparsely developed. There is just a plain huge amount of land with no one living in it, and another vast amount of land with very few people living in it. While on the other hand, there are very few Manhattans. There are more, but still very few, medium dense places somewhere in-between.
Some people want to live in more dense places, because they get the benefits of being near others packed more densely. While on the other hand, some people would prefer to live in less dense places, because they get the benefit of more land and space to themselves. And some prefer something in between.
Empirically, the demand for dense spaces is very high compared to the supply. And the demand slopes upwards with density, while supply slopes downwards. You can tell this is true because of the high prices of housing per sq ft in dense regions.
Whereas empirically, we are vastly oversupplied with distant rural zones where you can live far away from anyone else. You can tell this is true because of the low price of housing per sq ft in less dense regions.
This seems so natural to us that it's hard to imagine it being any other way. But that's just an artifact of having lots and lots and lots of land that is mostly empty.
Imagine living in a world where most of the land is super dense mega-packed city, far more than anyone can use, but where apartments are still plentiful as water. Where MegaManhattan goes on for miles and miles and miles and has so much housing that even though it's incredibly dense, there's still plenty of apartments vacant to move into.
And also imagine that relative to our world, there is a very small amount of empty land. Only a few places in this world where there is vast open space. Only 10,000 total homes exist, in this low density region.
Which is more expensive now: high density, or low density? Obviously in this other world the answer is low density. The demand for such low density housing would vastly outstrip the supply, whereas the plentiful apartments throughout MegaManhattan would ensure prices stay low.
There is nothing intrinsically expensive about high density regions, or low density regions. We just chronically underbuild high density by making it illegal to construct more, so prices stay high. If you make it impossible to build in a desirable low density region, you get sky high prices also (see: Aspen, CO, which is incredibly expensive vs its density).
I am not an economist, but I made an observation that got me thinking, and I came to a plausible explanation.
The observation: there are quite a few leftie environmentalists who insist that adding extra lanes to congested roads is pointless - the added supply will only induce more demand, so the road will stay just as congested.
Then they change the topic and demand that new, affordable housing is built in crowded cities. Hmmmmm.
So here's my simplified picture: smart, ambitious people move to big cities not because they want to, but because they have to, to find suitable jobs both for them and their equally highly-qualified spouse.
Big companies move to big cities because only there do they have a sufficient pool of smart, talented employees. Also, wealthy cities can offer good infrastucture and other incentives, and the wealthy bosses can shrug off the high costs of living and enjoy the cultural amenities.
Small companies move to big cities because that's where their clients are - the big companies and their well-paid employees.
Employees of small companies move there because it's still better to pay outrageously high rents than to be unemployed in a dead town.
All of this works the better, the more densely the city is packed (mostly in terms of "how many people can you reach with a 45-min ride in whatever means of transportation").
Nowhere in this argument did "it's fun to live in big, densely packed cities" enter the picture, except for the privileged few.
So as long as there are smallish cities and villages that the moloch monopoly city can suck dry, all that happens when you build another block of affordable housing in a big city is that the ratchet is cranked another notch toward "everyone lives in overcrowded, expensive urban hellscapes where only the richest few actually enjoy living".
How about we try to make smaller, less expensive cities more attractive to companies instead?
This sounds like a giant argument for a Land Value Tax. It is the only option which would increase building everywhere therefor decreasing home prices everywhere.
It should be very easy to convince you this is wrong:
So I don’t understand why Matt believes that building a few new apartments in some city - a very small move along that spectrum - would do anything other than make local prices go up
You are treating your model as though R2 is one. Your own data show its not. If there are cities both well above and well below trend line AND studies show small increases in supply controlled for other factors can reduce prices, then maybe take those results at face value. Or at least find data that show otherwise.
Theres no reason SD cant look at miamis dot and think, maybe we could get there with different policy decisions. It doesnt require a wierd divot in the curve.
If you want to be concerned that the larger scale relation cannot be improved, that is at least supported by your data. But movement along this curve is a long term change and multi factored and not strictly a function of how much upzoning happens in a given year
I think you’re conflating housing prices with amenities. Look at prices in Dallas vs Seattle. Seattle prices are higher than Dallas despite having fewer houses, because there is more demand, because there are more amenities such as restaurants, cultural activities, events, and high-paying jobs. This holds true even in small, expensive cities like Boulder, where there are popular amenities (beautiful outdoor activities, quaint downtown, etc), but not much housing. Large cities, by virtue of having more people, generally have more amenities. So there is a correlation, not causation.
To be clear, you may think that having more houses causes more people which cause more amenities, and it can be messy, but if that we’re purely the case we’d see no cities shrink while others grow, but we see that all the time.
Is induced demand is based more on the rank-order size of the city or the absolute size? I suspect the former is a big factor because there are dozens of large cities that don't even cross the mind of the sort of people who "would be happy to live in Seattle, or NYC, or the Bay". I count 56 metro areas above 1 million, and 15 metro areas above 4 million on this list: https://en.wikipedia.org/wiki/Metropolitan_statistical_area
There’s clearly a positive feedback loop causing urbanization. This is why cities exist in the first place and get larger and denser. (Contrast with farmers.)
The strength of the feedback loop varies by city and neighborhood. Some cities pull more people in and others don’t. We should expect higher-priced neighborhoods to have more of a draw for some reason.
Positive feedback loops can go on for quite a while but don’t necessarily go on forever. For example, demand for student housing from a university is limited depending on the growth rate of the university, and it might not be growing at all.
So I think this can’t be answered from first principles. You need to look at what’s driving demand locally. Will it continue?
I have long thought that dense California cities should be encouraging large employers to move, or at least expand more elsewhere. When an employer from a high-priced city moves to a city that needs more jobs, both places benefit.
"But Manhattan and London have the highest house prices in their respective countries, primarily because of their density and the opportunities density provides" implies a commutative relation that does not actually exist.
Yes, at the macro level, greater density -> more specialization -> more jobs/opportunities. But with transportation tech and cultural assumptions that commute time is the employee's problem, we have managed to severe the connection the other direction, such that we can create lots of opportunities/jobs/industry without also creating the resulting density.
I don't have the data handy to make plots, but there's a "natural" density of housing to match a given density of jobs within a metro area. A lot of our currently prosperous cities have made policy choices to prevent the housing density to grow to match the opportunity density. At that point, building more houses does *not* correspondingly increase the opportunities/make the city more desirable. It just enables the people in the existing jobs to be closer to them on average. If you built enough to actually start increasing opportunities again, then sure, prices will rise, but until then you aren't actually moving up that size/cost curve you started with.
I don't think you have justified excluding non-US metros, like Tokyo, or Auckland. Doesn't this lead to the natural conclusion that there is a sufficient level of housing to build, and that the problem is that the USA's many metros are structured to prevent housing? It seems like you're just arguing that US metros are bad at building housing, which is also what Matt Yglesias is arguing.
Your argument relies heavily on cherry-picked examples (New York City, San Francisco, North Dakota) and does not account for the broader housing market in the United States. This selective approach fails to consider the variety of factors that contribute to housing prices in different cities, and makes your argument less credible.
The assumption that density is the primary driver of high housing prices in cities like New York and San Francisco is simplistic and ignores the myriad of factors that contribute to their housing costs. Factors such as land scarcity, regulatory barriers, and high demand due to economic opportunities should not be overlooked.
The hypothetical scenario of Oakland becoming ten times denser is an extreme and unrealistic example. It fails to consider the gradual changes and adjustments that would occur in the housing market over time. Your argument does not account for the possibility that increased housing supply could eventually outpace demand, leading to stable or even lower housing prices.
Your claim that there is an inherent "coordination problem" between cities when it comes to upzoning and housing supply is unsubstantiated. You do not provide any evidence or examples to support this assertion, and it appears to be conjecture rather than grounded in economic theory or empirical evidence.
You fail to address the role of elasticities in the supply and demand relationship within housing markets. By not considering how elastic the demand for housing might be, your argument misses a key component in the analysis. If demand is relatively elastic, even small increases in housing supply could lead to lower housing prices.
Any way to analyze housing markets without considering the blatant and heavy handed manipulation of the real estate markets (either for gdp or property taxes) seems..... mildly optimistic about the price mechanism.
Mortgages are 30 year things and there's always some government program, theres going to be massive lag with unclear data at minimum.
Prices work best when the value of money is effectively linear, there are repeat costumers and products are interchangeable; which simply isn't the case for housing. You shouldn't assume that real estate has sane prices, and from deducing how a sane system would react to a change, conclude which effect.
My guess is there's a short term and a long term effect. In the short term building housing lowers house prices, as you'd expect.
In the long term more people move in to those houses, agglomeration effects kick in, and the city becomes more desirable. This raises house prices.
I think you are using ill-fitting metrics and it confuses you.
Prices are high where the supply is less than the demand. Absolute values of supply and demand are not relevant. So the prices/density graph is not helpful. Most of the causality there is reversed and the second order effects depend on variables for which density is itself a proxy: amount of amenities and opportunities.
If the amount of amenities and opportunities in a place is increased, demand for houses will rise, thus housing prices also rise, and until possible new houses are build, increasing density. If you just increase density without creating new jobs, nighclubs, museums and other places of interests that will motivate more people to want to live there, the prices will fall.
Let me rephrase the argument: positive network externalities to building densely in cities are so large that building more housing actually causes existing units to be so much more desirable that its price goes up, despite the additional supply.
This might be true (it's certainly logically possible), but I suspect it's false on the current margin in most housing markets, even really weird ones like Oakland or Manhattan. This seems to be what the empirical evidence says, as far as I know.
Notably, if this were true the cost of overly restrictive zoning is significantly greater than the consensus view.
One example of where building a huge amount likely lowered price is Chicago's far northern lakefront just south of Evanston, where a huge number of ~30 story apartment buildings were built along Lake Michigan in, I'm guessing, the 1960s. By the 1990s, condos in them seemed pretty cheap.
In general, though, the U.S. doesn't have that many places where developers were allowed to run hog wild building high rises. In contrast, Sao Paulo, Brazil struck me as having an immense number of residential high rises by American standards.
Southern California, for example, has only a tiny number of high rises along the Pacific.
Someone on here, or another Substack comment section, said that the recent revealed American preferences in housing outside NY, and maybe SF, was a house with 3-4 beds and a yard or garden. Cities with those kinds of housing stock are expanding, the rest (excluding the two I mentioned) are declining or static. Looking at population flows in the last few years that seems to be true. California has lost 500k people to outward migration in two years. Surely you would expect house prices to fall given those statistics particularly if the trend continues - 2.5M in a decade would be a major decline.
Given that new reality Scott is probably wrong about Oakland and SF. Oakland seems to be an overflow city, the place to be if you can’t afford SF. Building new housing there now into a population decline would see prices fall. SF on the other hand may take up some slack.
What percentage of the earth's surface within, say 50 miles of the Dallas city hall could be developed for housing? I'm guessing somewhere around 95%, leaving out some lakes and the like. In contrast, what percentage within 50 miles of the San Francisco city hall could be developed? I don't know, maybe 50%? Maybe less. A huge fraction is saltwater and quite a bit more is rugged mountains.
Not surprisingly, land costs more near San Francisco than near Dallas.
You have 2 assumptions:
1. density is desirable. This means that people mostly want to live and get higher amenities from a higher density city. I think this makes a lot of assumption on development that comes together with density - for example on public transportation, etc. But let's assume we get them with density.
2. There is a fixed amount of people willing to move to a metro in the US.
If you build more houses -> desirability increases -> each person in the city gets more from the city -> it attracts more people -> prices increase in the city, but reduce in the rest of the US metros.
Even if this is true, each person in the city gets more, so this is reasonable. Also, if this is done in all the cities (NY + Austin + SF + Boston + Auckland + Atlanta, etc) this will reduce the prices in all the US.
I spill 3 facts I know about urban economics and interesting top notch academic papers. I hope it helps.
(1) All else being the same, housing prices decrease when you build more (assuming same number of people, same wages, ...)
(2) Urban development increases productivity, wages and GDP/capita increases. (economies of agglomeration) As a consequence, wages increase and there are more jobs. More people move into cities. Commute times determine the limits of growths.
(3) Housing market equilibrium, and city growth is determined by multiple factors. Eventually negative externalizes will limit the growth. In the US housing markets, low density and bad public transport create negative externalities earlier than elsewhere. US economy would be much bigger if Urban planning and zoning would be better.
Two top notch relatively recent academic papers on the subject:
Hsieh, Chang-Tai, and Enrico Moretti. 2019. "Housing Constraints and Spatial Misallocation." American Economic Journal: Macroeconomics, 11 (2): 1-39. DOI: 10.1257/mac.20170388
Richard Hornbeck, Enrico Moretti; Estimating Who Benefits from Productivity Growth: Local and Distant Effects of City Productivity Growth on Wages, Rents, and Inequality. The Review of Economics and Statistics 2022; doi: https://doi.org/10.1162/rest_a_01208
I don't think the primary driver of demand is "wanting to live in a big city". I think the primary driver of demand is new york city wages. So if NYC built more housing the prices would drop as demand is met for all those who can achieve wages commensurate to the housing prices. Similarly if Oakland built 5 million houses they wouldn't see NYC style demand since there isn't NYC style wages
Maybe the relation between new housing construction and rent price go like this:
1) new housing -> rent price goes down (classic offer and demand)
2) more house -> new opportunity for services and shops, i.e. increase the productivity per m² of the place
3) increased productivity per m² -> more demand for housing ->rent price goes up (classic rent price depending on the place average salary)
So the apparent paradox is simply a 2 different effects with a time delay. New construction drive rents down in the short term, then up after some delay (if the city is succesfull, if not, it's a housing bubble crash)
Scott, I think you are missing the elephant in the room, which is jobs. If density is highly correlated (like the R^2 of 0.77 in your graph), Palo Alto should be cheaper than most US cities. The reason why it is not is because it is in close proximity to high quality, high paying jobs. On the other side, the reason why Detroit is doing bad for such a large city is due to the lack of jobs.
The causality is clearly: Jobs -> housing demand -> building. Yes, there are jobs that are created due to agglomeration effects, but for the most part, this agglomeration takes a long time. (Even Silicon Valley is built on top of 70 years of investments in tech). On the other hand, when oil is found in North Dakota, prices for small town housing rocket up.
Your thought experiment of Oakland was really weird, as you picked just a single city in a 10 million people metro area. There are enough jobs in the Bay Area that are already established for the area to absorb 4.5 million new residents. Applying your thought process to other cities shows the brokenness of the logic. If you magically turned Portland or Detroit into a 6.5 million person city overnight, there is no doubt that the price of housing would crater in Portland's case, and Detroit will definitely not become a new Manhattan.
If you accept that jobs create housing demand, then the solution is to keep increases in housing supply above increases in jobs. There are many ways of achieving this, some good, some bad. One would be to disallow cities from creating new office and industrial space without a corresponding increase in built housing.
If this is correct, then the intuition that Silicon Valley folks are increasing rent for everyone in SF is clear and logical, as they represent new jobs without the corresponding increase in housing.
> But it can’t do this forever - at some point, it will exhaust the pool of Americans who want to move to big cities (you’ll know this has happened when housing prices are no higher in big cities than anywhere else).
This is obviously wrong; most things are more expensive in cities than they are outside of cities for the very simple reason that people in cities have more money than other people do. It doesn't mean the urbanites want those things more; in many cases they want them less.
But even worse, you're treating a complex function of several variables, the price of housing... somewhere... as an indicator for a single variable, the size of "the pool of Americans who want to move to big cities". In the first place, that quantity is not a coherent concept: the number of people who want to move to cities is going to be affected by local conditions in the cities. If housing there gets cheap enough, more people will want to move than previously did. And in the second place, the general conceptual approach of judging the value of one variable by watching what happens to a complex composite figure that includes that variable is complete nonsense.
You are confusing density of houses with density of people.
If Oakland built 50,000 houses, that wouldn't make it attractive since people don't want to live in a town with 50,000 empty houses.
If Oakland built 50,000 houses and managed to fill 49,000 of them than that would make the remaining 1,000 more attractive since now Oakland is a bigger town with more amenities and culture. Although it takes time for amenities and culture to happen but yes, I would guess that if they built 50,000 new houses, managed to fill up 49,000 of them and promoted culture or waited for culture to happen naturally then the remaining 1,000 would be more attractive.
In terms of an "economic perpetual motion machine" this is not a big deal. I mean, any company that generates revenue and uses that to reinvest in the business is kind of like an "economic perpetual motion machine"
You could also do something like build 50,000 apartment units with great amenities, rent them out super cheaply for 5 years to make sure they get filled up and then once people moved in and culture and society is created jack up the prices. Don't know that anyone has ever tried this though since it sounds really risky. Building 50,000 apartment units is really expensive and you don't know if it will work. But the fact that "maybe there exist risky investment opportunities that if executed properly might yield massive returns over the course of a long period of time" doesn't seem to violate any kind of economic law does it?
Didn't Dutch originally settle in the area exactly because of the harbor? Wikipedia says
> The mouth of the Hudson River was selected as the ideal place for initial settlement as it had easy access to the ocean while also securing an ice-free lifeline to the beaver trading post near present-day Albany. Here, Indigenous hunters supplied them with pelts in exchange for European-made trade goods and wampum, which was soon being made by the Dutch on Long Island. In 1621, the Dutch West India Company was founded.
Surely geography helped New York later become a national capital and in any case a center of finance even after the establishment of DC.
You would obviously be wrong for any city in which demand is limited. If you found a city where nobody wanted to live, and put a lot of housing there, you would crash the price of houses in that city. You could use this idea to regulate housing prices in those kinds of cities - just build enough housing to lower prices to the level you desire, rather than crashing it. China has tried building cities from nothing, and many large buildings were later torn down having never been occupied. Even with as many people as China has, there are lots of places with existing housing available and nobody trying to live there.
The problem you seem to be identifying is that there are some cities where demand far outstrips supply, due to their nationwide pull. Lots of people want to live in NYC or SF who are unable to do so. This is coupled with the reality that developers are going to want to take advantage of that by building denser and more expensive housing, not cheaper and less dense housing. Because land is geographically isolated (you can't move the underlying land or the land next to it), there's only so much land available within a certain distance of whatever people want to be near. For SF it's likely tech jobs. Living a three hour drive away isn't useful if you want to work in tech on site. It's also not useful if you want to be a teacher or service employee in SF. Those could be great jobs, but you have to live close enough not to turn all your free time into a miserable commute.
I'm not sure if it's a counterexample or evidence for your theory, but Tokyo is notorious for being rather cheap for such a big, densely populated city. Their prices for houses in the 23 wards are about half of what you'd pay in the suburbs of Toronto (and Toronto in turn barely compares to Vancouver or SF).
On the one hand this could be explained away by the overall demand in Japan being capped. Tokyo is still the most expensive city in that country, and there's relatively few global migrants choosing between US cities and Japan ones.
On the other hand, Tokyo still has about 3x the population density of New York, so if they'd stuck to New York-style zoning, prices could have risen to absolutely insane levels (in the city itself and in whichever places people priced out chose instead). Conversely, New York moving to the density of Tokyo's metropolitan area (which is about the same size as New York itself), would support about 10,000,000 extra people. How big do you think the "any big city" pool of migrants is?
On the gripping hand, since the zoning rules are set nationally, it's not an example of "one city upzones drastically, but its neighbors don't". Maybe there would be significant extra demand if little-to-no housing supply could be built in Yokohama/Osaka/etc.
Find one of those little islands in Maine, then rebuild all of New York City on that isalnd in Maine. Are the new apartments in the new city worth as much as an apartment in NYC? Probably not because none of the people, businesses, institutions, etc in the original NYC that make it such a desirable place to live in are in Maine. There is a difference between population density and housing density. Population density makes housing prices go up. Additional housing make prices go down. Housing cost is the ratio of population density to housing density, ie the ratio of supply to demand like in any other market. So building additional housing shoud make prices go down in the short term, while long term you may have to deal with Jeavon's paradox whereby the increasing livability of the city makes it even more desirable to live in, causing an increase in prices, but that's not a given.
"No; I think the missing insight is that there’s some pool of geographically mobile Americans1 who are looking for new housing (or who might start looking if the right situation presented itself). These people have various combinations of preferences and requirements. One common pattern is to prefer any big city - they would be happy to live in Seattle, or NYC, or the Bay, if the opportunity came up."
Another common pattern is to want to live where you're from. In 1990 NYC had 7.3M people, compared to 8.8M now. How many of those 8.8M have lived here that entire time, or were born here during the interim? I was born and raised here, I like it here, and most of my friends and family are here. I'm definitely a big-city-simp, but Oakland would have a hard time getting me to move there by building 10X the housing for many of the same reasons that I don't currently move to Tokyo.
In the absence of zoning laws, housing costs will equal the cost of construction. If you get extremely dense like NYC and run out of land you'll need to build way up, which is very costly.
But most places in America can get 10x denser without having to build more than 3 stories high. If the area has housing prices higher than the cost of construction, eliminating zoning laws will lower housing prices even while inducing demand; you can just keep building to accommodate new arrivals!
As an analogy, we induced demand for cars by building lots of cars and roads. But cars didn't get more expensive, since we can scale up production to meet demand. Car prices will only increase if the cost of construction increases (albeit with short term fluctuations from sudden spike in demand).
Cities in the past were not this expensive for that reason. Chicago grew from 4,000 to 3.4M in less than a century, but managed to accommodate everyone.
I think more housing increases prices because it's limited. Looking at San Francisco, the city proper isn't even a million in population (the metro area is 7.7 million).
I have no idea how many people want to move to San Francisco, but let's say that it's 100,000 out of the entire world (because you'll get the best jobs in Silicon Valley and get rich).
If SF builds 90,000 new housing units, that means 90,000 more people can move there. So it's not going to bring down prices for those already living there, who might like to move out of a houseshare and get their own place, since the demand is continuing at the same level. And the price won't come down, because the housing supply isn't at the same level as housing demand - there are still more people who want to live in SF than there is housing available.
So increasing density won't bring down prices in that case. Unless you can say "In 2028 the projected population will be 2 million and by then, with old and new housing stock, we will have 3 million units available", then the price will come down. But so long as more people want to/need to move there, so that new housing is being snapped up as soon as it becomes available, then prices will remain high.
Same for the metro area - so you can't afford to pay rent to live in SF proper, then you move to the nearest city/town where you can pay rent. Soon everyone is taking that as their second choice, and demand is pushing up prices. Then people switch to their third choice, and so on, until you get to the places where the distance in travel time is so far, and there aren't commuter trains, that it's not in demand and so prices are lower.
I’m not entirely sure I disagree, but this also seems like an unnecessary addition that doesn’t explain much--particularly as it is contradicting by empirical studies.
Here’s what I see happening:
1) Land prices increase when an area is unusually productive or desirable.
2) This then incentivizes more housing units to be built to take advantage of that additional productivity/desirability.
3) Reactionaries (not necessarily a bad term here) oppose further construction, leading to continued increases in price.
There are two additional effects, one of which you posited
A) As density increases, the desirability of the city for urbanophiles increases.
B) As density increases, productivity increases because there are more talented people in a smaller radius and because you can tailor your local business to a smaller niche of people while having the same number of customers.
Your argument that ‘A’ is a main factor in driving prices up is fairly unconvincing to me. Plenty of people, if not most people, seem to prefer the suburbs to cities, at least aesthetically. On the other hand, ‘B’ seems fairly likely.
But what I want to specifically point out is that your argument rests on the claim that the cost increases from section 1 and 3 are smaller than the cost increases from section A (and B, if you include that).
Urban desirability and increased productivity (A and B) are positive feedback loops whereby increased housing leads to increased demand and therefore increased prices. However, ordinary supply and demand is a negative feedback loop whereby increased housing leads to increased supply and therefore decreased prices.
One important difference between these two models is that, if B is driving the increase, local wages should be expected to actually grow faster than local housing prices even if said housing prices increase, because the attracting of productive talent and productive companies (combined with Baumol’s cost-disease) will cause wage gains across the spectrum.
The only losers in this scenario are individuals wealthy enough that they would have preferred constant wages/salaries in exchange for continued suburbia.
The simple version of the argument is:
a) relaxed housing restrictions increases supply which decreases prices and rents
b) increased density creates more valuable urban agglomeration effects
c) urban agglomeration value is capitalised into valuation of dwellings, increasing demand and increasing prices
d) c) is bigger than a), so the net impact is that prices and rents rise.
The problem is that part d) cannot be inferred from the data (that would be) behind the imaginary chart because each of the cities on it have restrictions on dwellings that we aren't controlling for (and would have a hard time trying to work out exactly how to control for in any real-world study).
It's also not quite the right framing to think of a reduction in zoning restrictions as in increase in the supply of housing. It is really an increase in the elasticity of the supply of housing - by reducing absolute legal restrictions as well as the implicit (though sometimes more restrictive) barriers from the approvals process, the supply of housing becomes more responsive to changes in prices/rents.
The simple conclusion in theory (for instance In the Rosen-Roback model of spatial equilibrium) is that, whether urban agglomeration (which is effectively a kind of productivity) winds up in prices and rents depends on the elasticity of housing supply. Where the housing supply is perfectly inelastic (i.e. no price increase can induce the supply additional housing) all productivity increases go to house prices. Where the housing supply is perfectly elastic (i.e. an infinite amount of housing can be supplied at the going price) all productivity increases go to wages and population growth.
The imaginary graph (from the article) could be explained by the following stylised set of facts: historically freer planning, but urban development commencing at different times (and at different rates for random "natural" reasons) followed by much uniformly strict restrictions on redevelopment everywhere. The initial level of density (and thus urban agglomeration) is different in different places, but with similar levels of restrictions on new development (and additional densification) housing supply is similarly inelastic everywhere and so prices today are largely determined by density levels. This stylised history is consistent with frequent observations that, much of the development in big cities would be illegal today (e.g. https://www.nytimes.com/interactive/2016/05/19/upshot/forty-percent-of-manhattans-buildings-could-not-be-built-today.html)
The reality is likely a messy version of this (because restrictions are different in different cities and have been implemented at different times). But the important lens is that the relationship between prices and productivity (and thus density) depends on a relatively inelastic supply of houses - exactly what is being alleviated by zoning reform and relaxation.
The other insight that I think is missing from Yglesias' (and your) analysis here is that growth is multifaceted. Density doesn't increase for no reason; it increases as a knock-on result of growth in jobs, and amenities, and clubs, and service organizations, and sports teams, and etc. These things are really, really heavily coupled.
When cities go through periods of "growth" meaning the broad definition above, housing demand tends to grow faster than housing supply, which leads to things like above-market bidding wars and spiraling rents. *That* leads to density. Building stuff takes time; the density follows many months to multiple years *after* the increase in demand.
So, I disagree with "density increases local prices" per se; *growth* increases local prices, which prompts denser development. Where my YIMBYism comes in here is not that you can make a desirable city cheap with lots of new housing - sorry, not gonna happen - but rather that policies which restrict density should be avoided precisely because they guarantee that the price shocks that come with growth will be sharper and take longer to resolve.
Part of why this is all so important is that when price shocks do settle, they don't resolve into big price drops - price increases in housing are very, very sticky. Homeowners *hate* selling at a loss, and will avoid it like the plague. Landlords won't cut rents until vacancy rates become extreme. Policy has to be designed to keep the price jumps under control, because once they occur there's no path down that isn't super painful.
It sounds like this is just a matter of accounting for covariates. Housing density isn't the only thing that makes an urban environment valuable to live in. There need to be jobs, transportation options, food availability, basic services like plumbing and trash collection, a lack of crime, weatherproofing, public spaces, amenities, and so on, all of which we might lump into the attractiveness of a location.
If you look along the vector between the random plain in North Dakota and NYC, both the attractiveness and the housing supply increase, and the difference in price indicates that the attractiveness increases more than the housing supply does. But when it comes to discussing changes to housing policy in isolation, you're operating on a different vector that keeps the attractiveness more or less constant.
A simple graph might help, which I can describe in words: Put attractiveness on the vertical axis, and housing supply on the horizontal. Price is highest in the upper left, lowest in the lower right. NYC is above and to the right of ND, but it's more above than it is to the right, so along that diagonal, price increases as you go up and to the right. Housing policy debate is about approximately horizontal changes, and on that axis, price decreases as you go from left to right, i.e. increase the housing supply.
Scott, I'm tempted to say that you're not even wrong, but in an interesting way.
Your theory is that increasing housing density in places where people want to live increases prices, but that it does so in increasing the value of living in the city as a whole. If you're right, and I think maybe you are, then it doesn't make sense to say "We shouldn't build housing because although it will increase supply, allowing more people to live here and lowering prices, it will also make almost everyone in the city better off, therefore increasing demand to live in the city and increasing prices!" If you're right, we should build more housing both because it allows more people to live in the city and because it improves the lives of nearly everyone in the city - there's more housing AND each unit of housing is more valuable to the people living in it than it was before.
Quibble one: I don't see anything in your essay that suggests that the increased value effect would raise overall housing prices more than the increased supply effect would lower them. Do you think this is inevitable or even likely, and if so why?
Quibble two: Of course, you might still oppose YIMBYism because you were concerned that increased density would change the character of a city in undesirable ways, or you might be concerned that particular groups would lose out on the increase in value provided by more housing.
Land value tax would solve this.
The corollary would be that if you demolish housing in big cities, prices would fall. This doesn't seem quite right?
Adding housing only increases density if the land area is fixed. If new land is annexed for the new housing, available housing can increase with no change to density.
And the units in the first graph, density per unit area, make no sense to me.
There is a difference between house prices (which include the land under them and so dramatically increase as density increases) and housing rental prices (apartments/condos can be stacked to make use of the pricey land). That being said building up is harder than building on a flat plane, so I would expect the rental prices to increase relative to a flat plane (if somehow building a 20 story building was cheaper we’d see them built in the middle of nowhere more). So density does inherently raise rental prices somewhat because it requires building up, but artificial restrictions on building density will increase prices even more than they would be without them. I think it’s also right that density can increase demand for a particular city as some people want to live in with the extra amenities of density (and tolerate/enjoy the density itself) vs people who want to live in open country or more suburban areas. Japan allows more density and construction everywhere (plus falling population) so their housing is just a whole lot cheaper overall, but still should be more in Tokyo than in the country due to the height. As with many things in Japan, I suspect they are probably more at an overall equilibrium with regard to people distributing throughout the country and having the interior/exterior space, and density/amenities for their income vs the US which is much more dynamic, has a great many more building restrictions and inefficiencies, and has eye popping income increases from the tech and oil industries.
We also do things like requiring double stairs and have worse fire rescue capabilities, which increase the cost of building up vs building flat in the US relative to many other places. Property taxes/Georgism can significantly affect this as well, not sure how that differs between the US and Japan. There are also significant cultural differences in how income is used and wealth is saved, with the US using housing as the way to build a typical household’s wealth.
Didn't Robert Lucas point this out in simple terms like a hundred years ago when he noted that cities are valuable because of the social connections and the more people/housing units, adjusted for proximity, the more social connections, and therefore the more valuable the housing unit or the restaurant table.
I wrote a post trying to change Scott's mind about housing:
https://perpetualcall.substack.com/p/a-simple-model-of-housing-prices
I don't think your graph of price vs. density is as ridiculous as it looks at first glance.
You can very well make a model where each single added house decreases local prices by a small amount while having an overall positive correlation between houses and price in aggregate.
You just need some event that is more likely to happen with many houses that increases the price by a disproportionate amount, like a school, supermarket or public transport being built or a high profile business moving to the area. Or, simplyfying the model, an event that happens once a certain threshold of houses have been met. The resulting graph would go down with each single house but would also jump up at certain points, thus increasing the price in the long run.
Of course, this is a very YIMBY-favoring interpretation. Ignoring other factors, it would favor YIMBYism even on a geographically local scale from a community perspective: Prices (mostly) go down because there is more supply of equal quality and prices (mostly) go up because there's more quality of equal supply.
For inhabitants of big cities, who can actually afford it now, it is undesirable to decrease housing costs in big cities, because that leads to an increase of poorer people in their vicinity.
Considering cities, I think its not a function of big cities attract people. No, its jobs and amenities.
100K city no airport. no art museum. few high paying jobs
500k city - has airport, has art museum, has ethnic food, has a high tech business
NYC level - has stock exchange and banking industry able to support top class amenity, also other jobs, that attract other amenities. OOOH, HAKKA FOOD. OOH, AN OPERA AND A SPORTS TEAM.
You're not adding houses to a generic city with no special amenities. As the city gets bigger based on industry it also gets more amenities and becomes more attractive. OOH BALLET, OOOOH, ARTISAN PICKLES.
I haven't even considered cachet or prestige!!!!
But basically this will vault certain cities very high up for high high demand.
Adding a few more houses might not be able to satiate the pent up demand that is so high, these urban locations have people commuting for an hour on a train to work there.
I think you're onto something. But you should check out this paper by Ed Glaeser and Joseph Gyourko: https://pubs.aeaweb.org/doi/pdfplus/10.1257/jep.32.1.3. From Figure 3, housing prices in most cities are pretty close to the costs of construction in those cities. Deviations from this rule are cities where people want to move but there are natural or artificial barriers to housing construction.
Your story is based on the idea that density carries attractive amenities. Anecdotally, I think the amenity value of city density has been enhanced by recent technology. For example, technology has greatly eliminated the need to wait in stupid lines (at the bank, at the post office, etc.). And when you do have to wait in lines, it's not nearly so bad because your phone allows you to still be productive and/or entertained. Technology has also made it easier to navigate a dense city and to find all those niche little shops or clubs or farmers markets or whatever that make a dense city so cool.
The basis for your examination is flawed.
Strictly speaking - density is irrelevant. What matters is demand vs. supply.
The housing building advocates use this incorrectly: they presume perfectly matching demand vs. supply - thus increasing supply reduces demand effects. Wrong.
First of all, no city is standalone. San Francisco has the entire Bay Area as a suburb; New York has its other, non-Manhattan boroughs as well as New Jersey, Washington DC has Maryland and Virginia, etc etc. These suburbs/adjacent areas constitute both latent and unmet demand - while not everyone in those suburbs wants to live in the city, the reverse is not true at all: few people in the city want to live in the suburbs. The net demand effect of these suburbs is clearly skewed towards the city. In this respect: Yglesias is completely correct in that building 1 or a dozen apartment complexes will do nothing to materially affect the supply/demand curve.
The second problem is that there are material effects from social mores and government policy. Both San Francisco and New York City have property tax caps, for example. These caps serve to reduce turnover because anyone who has owned a property in those areas has a material tax benefit that increases over time - they only sell if they have to and/or are moving out forever. There are numerous properties in Pacific Heights in SF which are $4M to $7M in value but are paying under $3000 a year in property tax (use Redfin to look around); the identical properties, if purchased today, would be paying at least $40K to $80K a year in property taxes. New York city has only had the tax cap since 2011 but the effects are similar, if smaller. Social mores - Tokyo is a prime example. People simply live in smaller places. San Francisco and New York city also have rent control - another market distorting mechanism on top of (and directly related to) the property tax increase limits.
Lastly, cities are where rich people live. Yes, some live on huge ranches in Montana or in a "regular" house in Omaha, but the majority of rich people live in cities and/or have one or more properties in a city. Rich people = more expensive property.
I find Scott's argument compelling, but to me many of the comments seem to be arguing about tangential points. So I wanted to try to reiterate what I thought was the core of the argument, for what it's worth. Less eloquently than Scott, alas. And then I'll try to offer a slight extension.
In this simplified thought experiment, imagine all housing units are identical, so we abstract away from the effect of changing housing quality. And imagine employers' location choice and employees' location choice are governed by the same factors, so we can think of a representative person's preference.
The core assertion: suppose you build 1,000 new housing units in San Francisco. This will probably cause _more than_ 1,000 people to move to San Francisco (specifically, people who would not have moved to SF in the absence of this new construction), and so housing prices will probably go up.
I know it seems weird, but I actually find the assertion pretty compelling, because the value of the network is proportional to the number of nodes in the network. If you build 1,000 new housing units, and fill them with 1,000 new entrants, doing whatever cool things those new entrants are doing, does that make the city more or less attractive to the marginal new 1,001st person deciding between SF or LA? "More" does seem like it has to be the right answer, most of the time. So if someone was indifferent between SF and LA given (old population, old prices), then they must prefer SF with (new population, old prices), since it now has a superior network. And hence equilibrium demands that prices go up until indifference is again achieved.
If I had to think about what might be missing from this picture, the only thing that comes to mind is that the intertemporal behavior of the system might be important. Rather than building 1000 new units, consider what happens when you build the _first_ unit. The 1st new entrant, in the old steady state, did not come to SF. At (old pop., old price), they chose LA. And while you now have a new unit, you still have the old population. So to change the first person's mind, you have to offer a lower price.
So this suggests we should expect, if a lump of new housing appears, it should have a short term negative effect on prices, and a medium-to-long term positive effect on prices. Which also makes sense because new entrants to a city don't immediately start contributing to its network effects; they have to be integrated into the network first (i.e., the famous celebrity chef entrant has to open their restaurant, the wealthy angel investor has to become known to the entrepreneurial ecosystem, the artist has to open their gallery, and so on.)
One last generalization. If the short-term and medium-term effects of a new housing unit on house prices have opposite signs, then in a continuous system, the relationship between prices and building rates will probably be non-monotonic. Think of a differential equation where changes in prices are a function of (new entrants - new units), and where new entrants' rate of entry is a function of current prices and _lagged_ population (where the lag is proportional to the amount of time it takes for the average person debating between SF and LA to update their decision model to include the network value of recent new entrants.)
So you can have one steady state with no new units, no new entrants, and no price changes. Or another steady state with very slow building of new units and increasing prices (because the rate at which new entrants are integrated into the network exceeds the rate at which new units are built). Or a third steady state with fast building, where units are built faster than new entrants are integrated, and we observe decreasing prices.
This could conceivably make sense of both SF and Tokyo.
Plot price change since 2000, say, against population growth over that same period. It will be immediately clear that there are no shortage of slow-growing cities. Control for average wages and it will be immediate.
I live in Whitehorse, Yukon. The average home here costs 50% more than the average home in Houston Texas. Yukon is about as sparse as you can be. It's got more land than Germany and the whole population wouldn't sell out the stadium where the little league world series is played. Density we ain't got. NIMBYs and city planners we got in spades.