One thing that I think people sometimes have trouble with intuitively is that a "permission slip" system renders every other input meaningless when it comes to housing supply (assuming demand outstrips supply). The permission slip is the binding constraint, and nothing else matters. If you made it so that buildings were constructed with a wave of a magic wand, with no input cost, housing supply would still be a function of the number of permission slips issued each year, and the value of this technological advance would be fully captured by the people that own undeveloped/underdeveloped land + permission slips.
In practice, the other players in the game don't let the landowners capture all of the spoils from the permission slip system -- the union extracts a certain wage, the government imposes taxes (including stuff like inclusionary zoning, which functions as a tax), the consultants and lawyers impose rules so that they get a cut.
When people then do a bottom-up analysis of the cost of a home, it looks like all of the costs are really high and therefore the problem is multi-faceted, but the costs themselves are a function of the permission slip system. For example, the value of "land" (as you note, really land + permission slip) is by definition a residual calculation, the difference between the cost of building and the value of the finished asset. If you lower the cost of the other inputs, the cost of "land" will go up to fill the difference. But in real life, the other inputs are almost certainly also distorted by the process above.
We don't really encounter quota systems in the wild very much, so we never have the chance to develop very good intuitions around them. We are used to lower costs leading to lower prices to the consumer, but we don't usually consciously think about the transmission mechanism (competing firms lower prices to maintain share so that the consumer eventually captures all the benefits). The quota system disables the usual transmission mechanism.
I think this is also why people intuitively reject the idea that solving something as complex as the housing crisis is something as simple as zoning reform. But, the quota system is the bottleneck, and the *only* way to improve a system is to fix the bottleneck. That is, by definition, what a bottleneck is! It's like saying you can't fix a complicated supercomputer by simply plugging it in.
I thought he did a good job of covering that issue, and I agree with you that in a supply constrained market, construction costs don't matter. If the secondary market price for Taylor Swift tickets is $1000, increasing the face value on the ticket from 250 to 500 isn't going to have any impact on the market price. But I still found this discussion useful as it gives you some idea of what the competitive market equilibrium price would be.
Yes, definitely, I didn't say it very well, but you are correct that this article is very useful even though construction costs don't matter for determining the the current price - knowing real construction costs are the only way to quantify the loss to consumers.
I'm not sure why other inputs would be distorted by zoning. Some naterials can be imported from other areas if local prices would be higher.
Separately, however, we have evidence that just large increases in slips won't make significant price changes. When Rio enacted a policy that allowed for 36% more units city wide resulted in a price fall of 0.5%. And this result intuitively makes sense - residential prices will be tied to the value the location creates to a person. Jobs pay more in cities, so these benefits will be extracted in prices.
> I'm not sure why other inputs would be distorted by zoning. Some naterials can be imported from other areas if local prices would be higher.
Two channels:
1. What I alluded to in my comment is in practice (at least in the US) that input providers will demand to extract some of the surplus from developers by passing rules requiring consultants get paid off, above-market "prevailing wages" and so on. Developers will agree to this so they can get paid as well. These payoffs show up in the accounting as high input costs, but they derive entirely from the quota system.
2. This post shows that 50% of hard costs (and presumably 100% of soft costs) derive from the local cost of labor. Well, how is the market local cost of labor determined? In large part through the cost of housing (30% or much much more of total household expenses)! So it's recursive in part. So even if (1) isn't a factor, this factor will step in to inflate costs.
> Separately, however, we have evidence that just large increases in slips won't make significant price changes. When Rio enacted a policy that allowed for 36% more units city wide resulted in a price fall of 0.5%.
You can't analyze the effect of increased supply without finding a way to control for demand, because prices are a function of both supply *and* demand. You will always find cases where rents plummeted without an increase of supply simply by looking for cases where demand dropped (eg NY during Covid). Similarly you will find cases where rents didn't fall that much following an increase in supply, by identifying scenarios where demand went up a lot at the same time.
Pretty much every study that makes some attempt to control for demand through some instrument (eg looking at two neighboring areas in the same city) finds a huge effect of loosened supply. (Anyone that invests in real estate understands this intuitively too.) The intuition there is that rents/prices are simply a measure of scarcity, so more supply chasing the same number of people will cause rents to go down, at least until the market price of housing equals the replacement cost.
> And this result intuitively makes sense - residential prices will be tied to the value the location creates to a person. Jobs pay more in cities, so these benefits will be extracted in prices.
The point is that the intuition is wrong. Prices are a function of value and *scarcity*. You can think of a million things that are very valuable but not that scarce and thus not that expensive (water, food, iPhones). But there are other things that are both valuable and scarce that are expensive, like gold. All that is happening with the housing market is that we are injecting artificial scarcity, such that housing is expensive - lots of bidders and not enough homes. If we withdraw the bottleneck, housing prices will move back toward replacement cost, just as it is for any other good.
Regarding 2 - I don't think that's the causal direction (high cost cities cause high wages). No firms will set up in a high cost location if it could easily go in a low-cost location. The reason cities have high wages is because they are more productive. The reason why therefore the same price of labor in other sectors of the city goes up is due to opportunity cost (similar to the Baumol Effect). Workers could choose to do something else in the city which pays more and thus their bargaining position is stronger. They can get higher wages. This also explains why certain cities stopped being high cost over time - firms left after it was no longer productive for them to be there (or the product failed), and city house values collapsed, wages fell (no outside opportunities weakening bargaining).
And yes - 100% agreed that demand matters. But this where the intuition comes in. I went over a long example in one of my articles. The short version: in equilibrium (holding all else constant) you cannot have a city that has $100 in income and $20 of costs and a city with $50 in income and $20 of costs. People would move from the city with the lower income.
That's why demand for these high income urban areas is, to simplify, 'infinite' (New York has a population of 8-9mln, while US population is 300+mln which dwarfs NYC). That's why we don't see much of an effect on pricing when supply goes up. There is a lot of people ready to move in (note - this is still a good reason to build, as we have more people in a productive area). We cannot have arbitrage between cities - that is, after costs, an individual's welfare should be relatively similar (holding all else constant) regardless where they live.
In order to absorb this demand, we would probably need enormous supply increases in some of these cities, well beyond any feasible construction rates (as was put by another substacker - some things like income or population grow closer to exponential rates, while inputs such as infrastructure grow linearly). The other goods you mention are far easier to scale than housing.
The Covid example you mention is an interesting one. The reason for the 'demand' drop is that 'access' to the income an area provides was no longer valuable. Once Covid hit and remote work appeared, it turned out that many people were happy to leave cities since they could maintain income without staying in cities (again suggesting that the income is not caused by high local costs). A recent paper (https://www.aeaweb.org/articles?id=10.1257/app.20210190) found just that. Prices for housing across cities got closer, and prices within cities got closer too - there was less value in living closer to work since remote work reduced that value.
Yeah, but that argument would only make sense if all of the industries in a city were equally productive, and that's definitely not the case.
In practice you have different export industries in each city with wildly different productivity. If you have a high wage industry expand rapidly, and you have building restrictions, then rents go up, which pushes out an existing industry with lower wages. (Basically, Dutch Disease.) Like in SF, growth of the high wage tech industry combined with building restrictions ends up pushing out Wells Fargo and Schwab, who can't justifying pay the higher equilibrium wages, which is how tech workers have a place to live. (And yes, the landlords extract the productivity gains via high rents, since they control the scarce factor, housing.)
If you allow housing construction, you would expect lower productivity industries to stay in town, and probably capture some spillover benefits from being near the high productivity industries. You would also stop displacing people who might not want to leave.
It's important to observe that cities don't exactly have magical pixie dust that raises everyone's productivity, even if it can look that way in the data sometimes. There are some high productivity sectors within cities, but since those sectors themselves face finite demand and employ finite people, demand is not infinite.
You can see a demonstration of this in countries with fewer building restrictions. Tokyo has something like 14 million people (out of 125 million in Japan), which is big, but far from infinite.
- "Yeah, but that argument would only make sense if all of the industries in a city were equally productive, and that's definitely not the case. "
Yes - that's why we don't have certain industries in cities (tradeable goods), but have plenty of other services like child care/schools that have not had productivity increases. People in these jobs get paid a lot more, because their bargaining position driven by their outside option is high.
Regarding productivity, we have evidence that agglomeration results in productivity increases, which is why we see such specialization in cities (separately, something I wonder if new remote work technologies will make moot). Interestingly, Silicon Valley is such an example of high productivity due to laws. That area was not actually the first tech sector - the east coast was in Route 128. But 'non-competes' made it less productive and firms moved.
On the Tokyo example, it is worth noting that Japan or Tokyo has not experienced economic growth in over 20 years. I wouldn't have expected their rents or house prices to go up under these conditions. Moreover, their overall housing affordability is quite comparable to the high cost cities of the US (it is more affordable in Tokyo as percentage of income spent on housing+transportation is lower but not by much. Some of this discrepancy may be accounted for by future income growth).
Regardless, building in cites is good (if we assume agglomeration results in productivity increases).
Seems like if land is the binding input and the other inputs are in a competitive market, then land should be able to bargain higher shares over time, subject to the downward limit of the profit margin.
Sorry for spamming but I think there’s also a very fundamental flaw in the way that we consider land prices in this: the entire sample is biased to the point of uselessness by the selection effects of where *new* *single-family* housing is built.
If (even quite dense) SFH were being built in quantity deep within major metros that figure would be wildly different. But in addition to making in-fill development within cities very difficult, we’ve also made development of green space within existing suburbs very difficult.
So the entire sample is deeply skewed towards exurban land development, which elides the depths of the problem: not just what it costs to get houses built, but the impact on where and to what extent they *don't* get built.
I disagree. The 20% land cost heuristic is extremely useful in understanding the point at which it's economical to redevelop a plot of land. It helps illustrate the challenges of missing middle and frames policy discussions around what we can reasonably expect from smaller zoning increases (not a lot). And why ADU bills have been surprisingly effective, given that homeowners don't place much value on marginal land.
That the measurement comes from primarily green-field developments (which actually are getting built) isn't an issue. The idea that you want to understand the costs for homes that "don't get built" is non-sensical. It's the classic measuring the bullet holes of the planes that don't get shot down. If you want infill housing to get built, its economics need to look more like greenfield homes.
That was rather the point of my initial post? We can understand how prices for single-family, detached homes pencil out under the status quo but are missing a solid understanding of many classes of projects which simply don't occur often.
Perhaps you're correct and urban housing also needs land costs to make up no more than 20% of the final price. Perhaps not, given that generally higher incomes for urban transplants support higher final prices; we just don't know.
What sort of regulatory and land use changes would be required to make attached SFH (row or townhomes) pencil out more widely? What about attached top-bottom duplexes of the sort common in Canada? Mid-rise condominiums?
Your discussion of land prices here wandered into contested terrain in economics.
Broadly there are two views about land prices.
Classical economists such as Smith, Ricardo, J.S. Mill, and Marshall thought that physical realities gave land a monopolistic character, which caused land prices to equal the full value of using land, including specific locational advantages (e.g. proximity to markets). The idea was that competition for land would drive up its price to that level because labour/capital could move to that location to compete for land, but no more land could be moved to that location to compete for labour/capital.
That view was taken for granted for a very long time.
The alternative view is the new Chicago-school view that you cite. In this worldview, physical scarcity in many places is less important than regulatory scarcity, land prices can be meaningfully decomposed into the value of "permission slips" vis-a-vis value of location, and deregulating land use until permission slips are worthless would leave location value unchanged and therefore cause land prices (and house prices) to fall.
There are certain interesting arguments in the Chicago view, but the method to decompose the value of land into the value of "permission slips" and value of location is fundamentally flawed.
It rests on the idea that in a free/unregulated market there will be arbitrage wherein land is moved from low-value locations to high-value locations to create new plots of land, with this process continuing until all land has an average price equal to the current price of low-value slivers of land (the "marginal" land price as identified by hedonic pricing regressions).
From a "construction physics" point of view you might want to comment on the Chicago-school premise that land can be moved vs the classical premise that land cannot be moved...
(There's an alternative claim about why marginal land prices could become the new average price, which is that since all existing buildings have zero value as capital goods, we can continuously demolish them, redraw site boundaries, and rebuild them to optimal economic density, and thus in effect be continuously recompiling marginal slivers of land into useful-sized plots. The premise is equally flawed. In reality cities have a built form history and those buildings are uneconomic to remove.)
You can read about the problem with "permission slip" numbers here:
Murray, C. K. (2020). Marginal and average prices of land lots should not be equal: A critique of Glaeser and Gyourko’s method for identifying residential price effects of town planning regulations. Environment and Planning A: Economy and Space, 0308518X2094287. doi:10.1177/0308518x20942874
Another implication of the Chicago-school view of land and housing is that since "freeing up the supply side" by deregulating land use will cause land prices to fall via competition, rather than rise due to landowners gaining new rights of use for free, the many landowners who lobby for these policy changes are actually engaging in a catastrophic failure to know their own financial interest.
From a housing affordability perspective I wish this were true!
From a political economy perspective I think it more likely that the classical economists were correct and that vested interests are doing a great job promoting their own interests.
Phenomenal work, thanks for breaking this down! I found the part about developers gambling on being allowed to build more units than the land is initially zoned for interesting. How often is it the case that governing bodies are flexible in that way?
In my experience in Philadelphia it’s pretty close to 100% except when very well-connected neighbors throw a bitch-fit.
Developer buys blighted, obsolescent, or abandoned property, usually commercial or industrial, and proposes to do *something* with it. Their initial offer is 2-3X the density of the surrounding neighborhood. They go through the hearings and public input process, get savaged, dial back to 1.5-2x density, offer the local park $250k of capital improvement, the civic association gets agreement enough to say there’s a consensus, ignoring the loud hold-outs, and the zoning board greenlights it.
When a project doesn’t go through in the end, it’s almost always because the holdouts are wealthy or politically connected enough to start throwing grit into the gears in the form of lawsuits and/or personal intervention from politicians *after* this cycle is completed.
I'm a real estate appraiser in Madison, Wisconsin, which is currently experiencing a huge boom in apartment construction. A *lot* of land parcels are being developed at densities higher than "anyone would have expected" before the pandemic. OK I can't 100% say for sure that it's due to the pandemic, but the change over the past 10-15 years has been radical. The City of Madison used to say "we insist that you build fewer apartment units on this parcel than you've asked for," and now they're just as likely to say "isn't there a way to pack on a few more units?"
This is really interesting! I love your all of your analyses.
One thing I'm curious about - apparently 35% of people in America live in apartments. Apartments seem like a place where efficiency gains can really come into play (many repeated units) and where zoning causes huge issues. Do you have any thoughts on the spillover effects of zoning laws on apartments?
Another thing I'm curious about is the cost of labor with relation to density. Tokyo housing is much cheaper than NYC and this partially decreases labor costs. Do you think labor costs would go down if places became much denser?
Worth noting that Japan has a much higher degree of factory fabrication in the housing supply chain than most countries, a deliberate policy and business response to their economic woes of the 90s. I would expect Labour is a lower cost as a result.
I would argue that density per se doesn't relate directly to housing prices - a big thing that drives housing price variation is how efficiently supply is able to shift in response to increases in demand. It seems that Tokyo is good at that, in part because it's got looser rules for land use (including looser density caps), but allowing greater density in the places people are most interested in living is just one important means to that end. A dense city with a static housing supply, like LA, will be expensive.
The US really needs to get over single family detached housing. It is simply the most expensive way to house people period, in terms of hard, soft, and land costs. One of beefs with the NAHB study is that it is more or less exclusively about SFD.
I’d like to see some data that looks at the differences in automotive dependent SFD and whatever examples of walkable multi family can be found in the US. I noticed driveways in the pie chart, but the cost of a garage is lumped in a lot of those homes, as is all the site preparation and utilities for getting to each individual home.
It’s also interesting (at least in my neck of the woods) single family and multi family are almost like two entirely separate construction industries. The inability to flex between these styles seems relevant.
Lastly, the lack of detailed design in residential is what is holding it back. The sector will struggle to progress until it learns to do this.
Progressives and planners need to get over their totalitarian desires to force US citizens into Honest Worker Housing. Seriously, get over yourself. Your social planner pretensions aren’t helpful.
I agree that a certain kind of progressive exists who has a fetish for dense living arrangements due to perceived environmental benefits and they generally want to basically force those arrangements on everyone.
But it's also obvious, when viewed through the lens of pricing and demand, that there are plenty of families who would willingly live in a nice 3-bedroom condo in a building outside downtown if it weren't almost impossible to build such units. And a lot more families who would consider a 4-bedroom townhome within shouting distance of their town's walkable core and or a nice urban walkable commercial corridor, if they weren't also fairly rare. Those two types of housing (along with inner-suburb single-family detached housing, which simply cannot be scaled) are precisely the ones which have seen prices go insane over the last several decades.
The point, when it comes to land use, is to try to get the government's diktat slimmed down just a bit, so many different options can be constructed in many different jurisdictions. Let the market shake out the "means and methods" a bit more thoroughly once supply isn't capped.
“An electrician that costs $78 per hour to hire in Houston might cost almost $170 to hire in New York City.”
There’s a whole line of microeconomics papers waiting to be written on the feedback loop between land costs and labor costs in construction. It seems extraordinarily obvious that in locales with high housing cost pressures, hard costs are structurally significantly higher too. Vendors price rent into materials prices, labor costs more because they’re buying or renting locally, timelines stretch due to staging, delivery seats, inspection and feedback times.
Which begs the question: to what extent? What fraction of hard costs are themselves due to land use challenges?
Based on some of those pricing tables, do you think that buildings with more units could help reduce costs to effectively help scale some of that. For example driving down stucco or roofing costs since you need far fewer roofs in apartment buildings (1) than many stand alone buildings (many)
"As we’ve noted previously, modern buildings are fairly materially efficient, and there’s no obvious path for using substantially fewer materials that doesn’t come with significant tradeoffs."
It's important to note that there is a big difference between the amount of material that is in the building as designed, and the amount of material required to construct that building. Often a large amount of the material purchased is thrown out as scrap or is used to correct errors during construction, and it isn't required in theory for the building itself.
This is best seen in shipbuilding. Your previous post about shipbuilding:
showed a chart where a Japanese shipyard (IHI Aioi) using 65% as many materials to build the same ship as a US shipyard (Levingston- it had licensed the exact same ship design from them). Obviously the US ship did not contain many more materials, but the extra 35% of materials came from scrap and less efficient use. The practices of US shipyards during this period are very similar to the practices used in modern building construction, while the practices of Japanese shipyards (and other modern shipyards) are much different.
For electrical cables, this difference in shipbuilding methods can be seen on this site:
The traditional way is shown on the very last image (OC4f). The different types of cables are stored on a large rack and used as needed by the workers. But in the more efficient modern Japanese shipyards of the day, there was a second set of blueprints for construction, where everything was planned and controlled to the last millimeter. This included the length of every single cable required, with the extra required to cover the likely margin of error (this was reduced as techniques got better). They had computerized machines which received data on each cable required, cut each cable to the correct length, and coiled them to a specified diameter before labeling them. They were then organized into pallets to be placed in the room where the workers would need them, with the components that would be used first placed on top. This resulted in cable remnants on the order of 2 meters. For a traditionally built ship, or indeed any building as large as such a ship (since housing construction is similar to traditional methods), that would be unheard of. The cable length usage (and other aspects of construction) is not controlled to that level of accuracy- some material usage isn't controlled to any level of accuracy, as long as the finished component meets the requirements. I doubt we'll ever know how much cable is thrown out as remnants during construction, since it's not easy to measure, but it's likely a large fraction of cable used during construction.
The other part where this can be seen is in line heating, in these 2 posts:
There were 2 main benefits to line heating: mainly that the steel curved plates were much more controllable, and would not warp out of shape when they were welded. So the manufacturers could eliminate things like excess margins of material on each side, knowing there wouldn't be unexpected gaps between plates that would have to be bridged. They could also paint most of the plates before assembly knowing they wouldn't have to be bent back into shape later, cracking the paint and forcing them to paint it again. The traditional methods had this problem, and a large amount of excess material was needed to cover unexpected gaps, and more labor was also required to rework and paint plates after assembly. This is described in the NSRP Line Heating document as "redundant, corrective, or inherently unsafe and inefficient" work. The best analogy for this rework would be drywall mud: more than 1 coat of mud is not required to seal the wall; it's just required to smooth out any non-flat sections of wall. In other words, it is only corrective, meant to undo the inaccuracy in the initial installation of the drywall panels rather than add a new component to the house. If there was a way to mount drywall plates and apply just one coat of mud on the edges in a way that was guaranteed to be smooth and flat, there would be no need for more than 1 coat of mud on drywall. In fact, if the design were detailed enough that each drywall panel was specified with its locations for screws, holes for electrical outlets, etc., then the drywall could be primed and painted before installation except in those areas and on the edges. Of course today there is no way I know of to install drywall this accurately from the start, but that represents the material going towards corrective work rather than directly contributing towards the construction.
The other way that line heating helped was to eliminate the need for the large plate-forming jigs that were built for each type of plate and then thrown away as scrap. It also eliminated the wedges and clamps that were welded onto plates to bend them into shape, and then cut off afterward to be thrown out as scrap. The jigs remind me of frames for concrete pouring, and that probably is their best analogy for construction material use. Depending on the concrete form used, a lot of wood (or other material) can be used for frames which are thrown out after construction is finished. And there are other parts that may require temporary supports or jigs that are thrown out when a building is done.
In short, the amount of materials used in construction (and in shipbuilding prior to the 1980s) can be much larger than the materials actually required in the design of a building. Shipbuilding after WW2 seems to have gained an obsession towards accurate control over every aspect of manufacture, so that excess materials (and labor) can be almost eliminated. But construction uses methods similar to older methods of shipbuilding, so their material usage can be much higher.
This is such a well written article!! I’m a real estate investor in Cleveland Ohio who focuses on buying distressed properties to renovate and rent. In lower priced markets it’s not economically feasible to build houses, you’re better off just buying a distressed house and bringing it back to life. I think the biggest challenges I’ve seen/heard from developers and builders is all the red tape that most cities make you go through just to build something. There’s a big stigma around real estate investors being greedy and although that may not be entirely untrue the fact still remains that the reason why housing prices have remained high is because the supply is low and cities making the planning and permitting process difficult doesn’t help anyone.
While it would be good to reduce the costs of housing everywhere, aren't the real problems with housing affordability concentrated in metro areas where the price of land dominates the cost?
There's room for cost savings in single family homes through simplification of design and components: fewer corners, simplified rooflines, reduced ornamentation (baseboards and quarter-round), fewer kitchen and bath built-ins (oven/microwave/vanity), and simplified electrical (wireless switches). Some of these are customer preference which could be changed with the right price and marketing and some are required by code.
Interesting piece, thanks.
One thing that I think people sometimes have trouble with intuitively is that a "permission slip" system renders every other input meaningless when it comes to housing supply (assuming demand outstrips supply). The permission slip is the binding constraint, and nothing else matters. If you made it so that buildings were constructed with a wave of a magic wand, with no input cost, housing supply would still be a function of the number of permission slips issued each year, and the value of this technological advance would be fully captured by the people that own undeveloped/underdeveloped land + permission slips.
In practice, the other players in the game don't let the landowners capture all of the spoils from the permission slip system -- the union extracts a certain wage, the government imposes taxes (including stuff like inclusionary zoning, which functions as a tax), the consultants and lawyers impose rules so that they get a cut.
When people then do a bottom-up analysis of the cost of a home, it looks like all of the costs are really high and therefore the problem is multi-faceted, but the costs themselves are a function of the permission slip system. For example, the value of "land" (as you note, really land + permission slip) is by definition a residual calculation, the difference between the cost of building and the value of the finished asset. If you lower the cost of the other inputs, the cost of "land" will go up to fill the difference. But in real life, the other inputs are almost certainly also distorted by the process above.
We don't really encounter quota systems in the wild very much, so we never have the chance to develop very good intuitions around them. We are used to lower costs leading to lower prices to the consumer, but we don't usually consciously think about the transmission mechanism (competing firms lower prices to maintain share so that the consumer eventually captures all the benefits). The quota system disables the usual transmission mechanism.
I think this is also why people intuitively reject the idea that solving something as complex as the housing crisis is something as simple as zoning reform. But, the quota system is the bottleneck, and the *only* way to improve a system is to fix the bottleneck. That is, by definition, what a bottleneck is! It's like saying you can't fix a complicated supercomputer by simply plugging it in.
I thought he did a good job of covering that issue, and I agree with you that in a supply constrained market, construction costs don't matter. If the secondary market price for Taylor Swift tickets is $1000, increasing the face value on the ticket from 250 to 500 isn't going to have any impact on the market price. But I still found this discussion useful as it gives you some idea of what the competitive market equilibrium price would be.
Yes, definitely, I didn't say it very well, but you are correct that this article is very useful even though construction costs don't matter for determining the the current price - knowing real construction costs are the only way to quantify the loss to consumers.
I'm not sure why other inputs would be distorted by zoning. Some naterials can be imported from other areas if local prices would be higher.
Separately, however, we have evidence that just large increases in slips won't make significant price changes. When Rio enacted a policy that allowed for 36% more units city wide resulted in a price fall of 0.5%. And this result intuitively makes sense - residential prices will be tied to the value the location creates to a person. Jobs pay more in cities, so these benefits will be extracted in prices.
> I'm not sure why other inputs would be distorted by zoning. Some naterials can be imported from other areas if local prices would be higher.
Two channels:
1. What I alluded to in my comment is in practice (at least in the US) that input providers will demand to extract some of the surplus from developers by passing rules requiring consultants get paid off, above-market "prevailing wages" and so on. Developers will agree to this so they can get paid as well. These payoffs show up in the accounting as high input costs, but they derive entirely from the quota system.
2. This post shows that 50% of hard costs (and presumably 100% of soft costs) derive from the local cost of labor. Well, how is the market local cost of labor determined? In large part through the cost of housing (30% or much much more of total household expenses)! So it's recursive in part. So even if (1) isn't a factor, this factor will step in to inflate costs.
> Separately, however, we have evidence that just large increases in slips won't make significant price changes. When Rio enacted a policy that allowed for 36% more units city wide resulted in a price fall of 0.5%.
You can't analyze the effect of increased supply without finding a way to control for demand, because prices are a function of both supply *and* demand. You will always find cases where rents plummeted without an increase of supply simply by looking for cases where demand dropped (eg NY during Covid). Similarly you will find cases where rents didn't fall that much following an increase in supply, by identifying scenarios where demand went up a lot at the same time.
Pretty much every study that makes some attempt to control for demand through some instrument (eg looking at two neighboring areas in the same city) finds a huge effect of loosened supply. (Anyone that invests in real estate understands this intuitively too.) The intuition there is that rents/prices are simply a measure of scarcity, so more supply chasing the same number of people will cause rents to go down, at least until the market price of housing equals the replacement cost.
> And this result intuitively makes sense - residential prices will be tied to the value the location creates to a person. Jobs pay more in cities, so these benefits will be extracted in prices.
The point is that the intuition is wrong. Prices are a function of value and *scarcity*. You can think of a million things that are very valuable but not that scarce and thus not that expensive (water, food, iPhones). But there are other things that are both valuable and scarce that are expensive, like gold. All that is happening with the housing market is that we are injecting artificial scarcity, such that housing is expensive - lots of bidders and not enough homes. If we withdraw the bottleneck, housing prices will move back toward replacement cost, just as it is for any other good.
Worth looking at Auckland (my home) as one if the most recent examples of what releasing the zoning bottleneck can achieve.
https://open.substack.com/pub/apricitas/p/new-zealands-building-boomand-what
Regarding 2 - I don't think that's the causal direction (high cost cities cause high wages). No firms will set up in a high cost location if it could easily go in a low-cost location. The reason cities have high wages is because they are more productive. The reason why therefore the same price of labor in other sectors of the city goes up is due to opportunity cost (similar to the Baumol Effect). Workers could choose to do something else in the city which pays more and thus their bargaining position is stronger. They can get higher wages. This also explains why certain cities stopped being high cost over time - firms left after it was no longer productive for them to be there (or the product failed), and city house values collapsed, wages fell (no outside opportunities weakening bargaining).
And yes - 100% agreed that demand matters. But this where the intuition comes in. I went over a long example in one of my articles. The short version: in equilibrium (holding all else constant) you cannot have a city that has $100 in income and $20 of costs and a city with $50 in income and $20 of costs. People would move from the city with the lower income.
That's why demand for these high income urban areas is, to simplify, 'infinite' (New York has a population of 8-9mln, while US population is 300+mln which dwarfs NYC). That's why we don't see much of an effect on pricing when supply goes up. There is a lot of people ready to move in (note - this is still a good reason to build, as we have more people in a productive area). We cannot have arbitrage between cities - that is, after costs, an individual's welfare should be relatively similar (holding all else constant) regardless where they live.
In order to absorb this demand, we would probably need enormous supply increases in some of these cities, well beyond any feasible construction rates (as was put by another substacker - some things like income or population grow closer to exponential rates, while inputs such as infrastructure grow linearly). The other goods you mention are far easier to scale than housing.
The Covid example you mention is an interesting one. The reason for the 'demand' drop is that 'access' to the income an area provides was no longer valuable. Once Covid hit and remote work appeared, it turned out that many people were happy to leave cities since they could maintain income without staying in cities (again suggesting that the income is not caused by high local costs). A recent paper (https://www.aeaweb.org/articles?id=10.1257/app.20210190) found just that. Prices for housing across cities got closer, and prices within cities got closer too - there was less value in living closer to work since remote work reduced that value.
Yeah, but that argument would only make sense if all of the industries in a city were equally productive, and that's definitely not the case.
In practice you have different export industries in each city with wildly different productivity. If you have a high wage industry expand rapidly, and you have building restrictions, then rents go up, which pushes out an existing industry with lower wages. (Basically, Dutch Disease.) Like in SF, growth of the high wage tech industry combined with building restrictions ends up pushing out Wells Fargo and Schwab, who can't justifying pay the higher equilibrium wages, which is how tech workers have a place to live. (And yes, the landlords extract the productivity gains via high rents, since they control the scarce factor, housing.)
If you allow housing construction, you would expect lower productivity industries to stay in town, and probably capture some spillover benefits from being near the high productivity industries. You would also stop displacing people who might not want to leave.
It's important to observe that cities don't exactly have magical pixie dust that raises everyone's productivity, even if it can look that way in the data sometimes. There are some high productivity sectors within cities, but since those sectors themselves face finite demand and employ finite people, demand is not infinite.
You can see a demonstration of this in countries with fewer building restrictions. Tokyo has something like 14 million people (out of 125 million in Japan), which is big, but far from infinite.
- "Yeah, but that argument would only make sense if all of the industries in a city were equally productive, and that's definitely not the case. "
Yes - that's why we don't have certain industries in cities (tradeable goods), but have plenty of other services like child care/schools that have not had productivity increases. People in these jobs get paid a lot more, because their bargaining position driven by their outside option is high.
Regarding productivity, we have evidence that agglomeration results in productivity increases, which is why we see such specialization in cities (separately, something I wonder if new remote work technologies will make moot). Interestingly, Silicon Valley is such an example of high productivity due to laws. That area was not actually the first tech sector - the east coast was in Route 128. But 'non-competes' made it less productive and firms moved.
On the Tokyo example, it is worth noting that Japan or Tokyo has not experienced economic growth in over 20 years. I wouldn't have expected their rents or house prices to go up under these conditions. Moreover, their overall housing affordability is quite comparable to the high cost cities of the US (it is more affordable in Tokyo as percentage of income spent on housing+transportation is lower but not by much. Some of this discrepancy may be accounted for by future income growth).
Regardless, building in cites is good (if we assume agglomeration results in productivity increases).
Correction - Sao Paulo, not Rio.
Seems like if land is the binding input and the other inputs are in a competitive market, then land should be able to bargain higher shares over time, subject to the downward limit of the profit margin.
Sorry for spamming but I think there’s also a very fundamental flaw in the way that we consider land prices in this: the entire sample is biased to the point of uselessness by the selection effects of where *new* *single-family* housing is built.
If (even quite dense) SFH were being built in quantity deep within major metros that figure would be wildly different. But in addition to making in-fill development within cities very difficult, we’ve also made development of green space within existing suburbs very difficult.
So the entire sample is deeply skewed towards exurban land development, which elides the depths of the problem: not just what it costs to get houses built, but the impact on where and to what extent they *don't* get built.
+1 New single family homes are an unrepresentative sample understanding general land costs of places that could be built but aren’t.
I disagree. The 20% land cost heuristic is extremely useful in understanding the point at which it's economical to redevelop a plot of land. It helps illustrate the challenges of missing middle and frames policy discussions around what we can reasonably expect from smaller zoning increases (not a lot). And why ADU bills have been surprisingly effective, given that homeowners don't place much value on marginal land.
That the measurement comes from primarily green-field developments (which actually are getting built) isn't an issue. The idea that you want to understand the costs for homes that "don't get built" is non-sensical. It's the classic measuring the bullet holes of the planes that don't get shot down. If you want infill housing to get built, its economics need to look more like greenfield homes.
That was rather the point of my initial post? We can understand how prices for single-family, detached homes pencil out under the status quo but are missing a solid understanding of many classes of projects which simply don't occur often.
Perhaps you're correct and urban housing also needs land costs to make up no more than 20% of the final price. Perhaps not, given that generally higher incomes for urban transplants support higher final prices; we just don't know.
What sort of regulatory and land use changes would be required to make attached SFH (row or townhomes) pencil out more widely? What about attached top-bottom duplexes of the sort common in Canada? Mid-rise condominiums?
I enjoy your detailed analysis.
Your discussion of land prices here wandered into contested terrain in economics.
Broadly there are two views about land prices.
Classical economists such as Smith, Ricardo, J.S. Mill, and Marshall thought that physical realities gave land a monopolistic character, which caused land prices to equal the full value of using land, including specific locational advantages (e.g. proximity to markets). The idea was that competition for land would drive up its price to that level because labour/capital could move to that location to compete for land, but no more land could be moved to that location to compete for labour/capital.
That view was taken for granted for a very long time.
The alternative view is the new Chicago-school view that you cite. In this worldview, physical scarcity in many places is less important than regulatory scarcity, land prices can be meaningfully decomposed into the value of "permission slips" vis-a-vis value of location, and deregulating land use until permission slips are worthless would leave location value unchanged and therefore cause land prices (and house prices) to fall.
There are certain interesting arguments in the Chicago view, but the method to decompose the value of land into the value of "permission slips" and value of location is fundamentally flawed.
It rests on the idea that in a free/unregulated market there will be arbitrage wherein land is moved from low-value locations to high-value locations to create new plots of land, with this process continuing until all land has an average price equal to the current price of low-value slivers of land (the "marginal" land price as identified by hedonic pricing regressions).
From a "construction physics" point of view you might want to comment on the Chicago-school premise that land can be moved vs the classical premise that land cannot be moved...
(There's an alternative claim about why marginal land prices could become the new average price, which is that since all existing buildings have zero value as capital goods, we can continuously demolish them, redraw site boundaries, and rebuild them to optimal economic density, and thus in effect be continuously recompiling marginal slivers of land into useful-sized plots. The premise is equally flawed. In reality cities have a built form history and those buildings are uneconomic to remove.)
You can read about the problem with "permission slip" numbers here:
Murray, C. K. (2020). Marginal and average prices of land lots should not be equal: A critique of Glaeser and Gyourko’s method for identifying residential price effects of town planning regulations. Environment and Planning A: Economy and Space, 0308518X2094287. doi:10.1177/0308518x20942874
Another implication of the Chicago-school view of land and housing is that since "freeing up the supply side" by deregulating land use will cause land prices to fall via competition, rather than rise due to landowners gaining new rights of use for free, the many landowners who lobby for these policy changes are actually engaging in a catastrophic failure to know their own financial interest.
From a housing affordability perspective I wish this were true!
From a political economy perspective I think it more likely that the classical economists were correct and that vested interests are doing a great job promoting their own interests.
Phenomenal work, thanks for breaking this down! I found the part about developers gambling on being allowed to build more units than the land is initially zoned for interesting. How often is it the case that governing bodies are flexible in that way?
In my experience in Philadelphia it’s pretty close to 100% except when very well-connected neighbors throw a bitch-fit.
Developer buys blighted, obsolescent, or abandoned property, usually commercial or industrial, and proposes to do *something* with it. Their initial offer is 2-3X the density of the surrounding neighborhood. They go through the hearings and public input process, get savaged, dial back to 1.5-2x density, offer the local park $250k of capital improvement, the civic association gets agreement enough to say there’s a consensus, ignoring the loud hold-outs, and the zoning board greenlights it.
When a project doesn’t go through in the end, it’s almost always because the holdouts are wealthy or politically connected enough to start throwing grit into the gears in the form of lawsuits and/or personal intervention from politicians *after* this cycle is completed.
I'm a real estate appraiser in Madison, Wisconsin, which is currently experiencing a huge boom in apartment construction. A *lot* of land parcels are being developed at densities higher than "anyone would have expected" before the pandemic. OK I can't 100% say for sure that it's due to the pandemic, but the change over the past 10-15 years has been radical. The City of Madison used to say "we insist that you build fewer apartment units on this parcel than you've asked for," and now they're just as likely to say "isn't there a way to pack on a few more units?"
This is really interesting! I love your all of your analyses.
One thing I'm curious about - apparently 35% of people in America live in apartments. Apartments seem like a place where efficiency gains can really come into play (many repeated units) and where zoning causes huge issues. Do you have any thoughts on the spillover effects of zoning laws on apartments?
Another thing I'm curious about is the cost of labor with relation to density. Tokyo housing is much cheaper than NYC and this partially decreases labor costs. Do you think labor costs would go down if places became much denser?
Worth noting that Japan has a much higher degree of factory fabrication in the housing supply chain than most countries, a deliberate policy and business response to their economic woes of the 90s. I would expect Labour is a lower cost as a result.
A few years ago I attempted an article like this about apartments. It also found that land cost accounted for around 20% of new home prices. https://www.sightline.org/2018/08/30/what-makes-portlands-new-apartments-so-expensive/
I would argue that density per se doesn't relate directly to housing prices - a big thing that drives housing price variation is how efficiently supply is able to shift in response to increases in demand. It seems that Tokyo is good at that, in part because it's got looser rules for land use (including looser density caps), but allowing greater density in the places people are most interested in living is just one important means to that end. A dense city with a static housing supply, like LA, will be expensive.
Thanks, I meant supply not density. Good catch and thanks for the pointer!
The US really needs to get over single family detached housing. It is simply the most expensive way to house people period, in terms of hard, soft, and land costs. One of beefs with the NAHB study is that it is more or less exclusively about SFD.
I’d like to see some data that looks at the differences in automotive dependent SFD and whatever examples of walkable multi family can be found in the US. I noticed driveways in the pie chart, but the cost of a garage is lumped in a lot of those homes, as is all the site preparation and utilities for getting to each individual home.
It’s also interesting (at least in my neck of the woods) single family and multi family are almost like two entirely separate construction industries. The inability to flex between these styles seems relevant.
Lastly, the lack of detailed design in residential is what is holding it back. The sector will struggle to progress until it learns to do this.
Progressives and planners need to get over their totalitarian desires to force US citizens into Honest Worker Housing. Seriously, get over yourself. Your social planner pretensions aren’t helpful.
I agree that a certain kind of progressive exists who has a fetish for dense living arrangements due to perceived environmental benefits and they generally want to basically force those arrangements on everyone.
But it's also obvious, when viewed through the lens of pricing and demand, that there are plenty of families who would willingly live in a nice 3-bedroom condo in a building outside downtown if it weren't almost impossible to build such units. And a lot more families who would consider a 4-bedroom townhome within shouting distance of their town's walkable core and or a nice urban walkable commercial corridor, if they weren't also fairly rare. Those two types of housing (along with inner-suburb single-family detached housing, which simply cannot be scaled) are precisely the ones which have seen prices go insane over the last several decades.
The point, when it comes to land use, is to try to get the government's diktat slimmed down just a bit, so many different options can be constructed in many different jurisdictions. Let the market shake out the "means and methods" a bit more thoroughly once supply isn't capped.
“An electrician that costs $78 per hour to hire in Houston might cost almost $170 to hire in New York City.”
There’s a whole line of microeconomics papers waiting to be written on the feedback loop between land costs and labor costs in construction. It seems extraordinarily obvious that in locales with high housing cost pressures, hard costs are structurally significantly higher too. Vendors price rent into materials prices, labor costs more because they’re buying or renting locally, timelines stretch due to staging, delivery seats, inspection and feedback times.
Which begs the question: to what extent? What fraction of hard costs are themselves due to land use challenges?
Take a look at this paper "No Price Like Home: Global House Prices, 1870-2012".
https://www.aeaweb.org/articles?id=10.1257/aer.20150501
The authors show that 80 per cent of the rise in prices in real esate prices in the last more than 100 years is due to the rise of land prices.
Based on some of those pricing tables, do you think that buildings with more units could help reduce costs to effectively help scale some of that. For example driving down stucco or roofing costs since you need far fewer roofs in apartment buildings (1) than many stand alone buildings (many)
I think it does, though there are other costs too of course. I attempted a version of this article for multifamily buildings several years ago, if you're interested: https://www.sightline.org/2018/08/30/what-makes-portlands-new-apartments-so-expensive/
Fascinating
"As we’ve noted previously, modern buildings are fairly materially efficient, and there’s no obvious path for using substantially fewer materials that doesn’t come with significant tradeoffs."
It's important to note that there is a big difference between the amount of material that is in the building as designed, and the amount of material required to construct that building. Often a large amount of the material purchased is thrown out as scrap or is used to correct errors during construction, and it isn't required in theory for the building itself.
This is best seen in shipbuilding. Your previous post about shipbuilding:
https://www.construction-physics.com/p/lessons-from-shipbuilding-productivity-4b9
showed a chart where a Japanese shipyard (IHI Aioi) using 65% as many materials to build the same ship as a US shipyard (Levingston- it had licensed the exact same ship design from them). Obviously the US ship did not contain many more materials, but the extra 35% of materials came from scrap and less efficient use. The practices of US shipyards during this period are very similar to the practices used in modern building construction, while the practices of Japanese shipyards (and other modern shipyards) are much different.
For electrical cables, this difference in shipbuilding methods can be seen on this site:
https://lou.chirillo.com/shipbuilding-database/outfit-components/
The traditional way is shown on the very last image (OC4f). The different types of cables are stored on a large rack and used as needed by the workers. But in the more efficient modern Japanese shipyards of the day, there was a second set of blueprints for construction, where everything was planned and controlled to the last millimeter. This included the length of every single cable required, with the extra required to cover the likely margin of error (this was reduced as techniques got better). They had computerized machines which received data on each cable required, cut each cable to the correct length, and coiled them to a specified diameter before labeling them. They were then organized into pallets to be placed in the room where the workers would need them, with the components that would be used first placed on top. This resulted in cable remnants on the order of 2 meters. For a traditionally built ship, or indeed any building as large as such a ship (since housing construction is similar to traditional methods), that would be unheard of. The cable length usage (and other aspects of construction) is not controlled to that level of accuracy- some material usage isn't controlled to any level of accuracy, as long as the finished component meets the requirements. I doubt we'll ever know how much cable is thrown out as remnants during construction, since it's not easy to measure, but it's likely a large fraction of cable used during construction.
The other part where this can be seen is in line heating, in these 2 posts:
https://www.alternatehistory.com/forum/threads/alternate-warships-of-nations.326948/page-367#post-23354483
https://www.alternatehistory.com/forum/threads/alternate-warships-of-nations.326948/page-367#post-23356384
There were 2 main benefits to line heating: mainly that the steel curved plates were much more controllable, and would not warp out of shape when they were welded. So the manufacturers could eliminate things like excess margins of material on each side, knowing there wouldn't be unexpected gaps between plates that would have to be bridged. They could also paint most of the plates before assembly knowing they wouldn't have to be bent back into shape later, cracking the paint and forcing them to paint it again. The traditional methods had this problem, and a large amount of excess material was needed to cover unexpected gaps, and more labor was also required to rework and paint plates after assembly. This is described in the NSRP Line Heating document as "redundant, corrective, or inherently unsafe and inefficient" work. The best analogy for this rework would be drywall mud: more than 1 coat of mud is not required to seal the wall; it's just required to smooth out any non-flat sections of wall. In other words, it is only corrective, meant to undo the inaccuracy in the initial installation of the drywall panels rather than add a new component to the house. If there was a way to mount drywall plates and apply just one coat of mud on the edges in a way that was guaranteed to be smooth and flat, there would be no need for more than 1 coat of mud on drywall. In fact, if the design were detailed enough that each drywall panel was specified with its locations for screws, holes for electrical outlets, etc., then the drywall could be primed and painted before installation except in those areas and on the edges. Of course today there is no way I know of to install drywall this accurately from the start, but that represents the material going towards corrective work rather than directly contributing towards the construction.
The other way that line heating helped was to eliminate the need for the large plate-forming jigs that were built for each type of plate and then thrown away as scrap. It also eliminated the wedges and clamps that were welded onto plates to bend them into shape, and then cut off afterward to be thrown out as scrap. The jigs remind me of frames for concrete pouring, and that probably is their best analogy for construction material use. Depending on the concrete form used, a lot of wood (or other material) can be used for frames which are thrown out after construction is finished. And there are other parts that may require temporary supports or jigs that are thrown out when a building is done.
In short, the amount of materials used in construction (and in shipbuilding prior to the 1980s) can be much larger than the materials actually required in the design of a building. Shipbuilding after WW2 seems to have gained an obsession towards accurate control over every aspect of manufacture, so that excess materials (and labor) can be almost eliminated. But construction uses methods similar to older methods of shipbuilding, so their material usage can be much higher.
Excellent analysis as always.
This is such a well written article!! I’m a real estate investor in Cleveland Ohio who focuses on buying distressed properties to renovate and rent. In lower priced markets it’s not economically feasible to build houses, you’re better off just buying a distressed house and bringing it back to life. I think the biggest challenges I’ve seen/heard from developers and builders is all the red tape that most cities make you go through just to build something. There’s a big stigma around real estate investors being greedy and although that may not be entirely untrue the fact still remains that the reason why housing prices have remained high is because the supply is low and cities making the planning and permitting process difficult doesn’t help anyone.
While it would be good to reduce the costs of housing everywhere, aren't the real problems with housing affordability concentrated in metro areas where the price of land dominates the cost?
There's room for cost savings in single family homes through simplification of design and components: fewer corners, simplified rooflines, reduced ornamentation (baseboards and quarter-round), fewer kitchen and bath built-ins (oven/microwave/vanity), and simplified electrical (wireless switches). Some of these are customer preference which could be changed with the right price and marketing and some are required by code.
An excellent analysis. I like how you deal with both geographical variation and the difference between new housing construction and existing homes.
So it really does all come back to NIMBY-ism. :)