It’s long been believed the constantly rising costs of new home construction, and lackluster improvements in construction productivity more generally, are fundamentally a problem of production methods.
Sweden and Finland have huge forestry sectors, with many local prefab industries having evolved near sawmills to capture more of the economic value. With the huge efficiency gains in forestry there has been a surplus of labor and cheap production facilities in these more rural areas.
It is also a mature industry. My parents prefab manufacturer was established 80 years ago. I’d say there are 10 or so strong brands that capture 50% of the market for single-family homes. That is were you start looking if you think of building a home. Different brands have different constructions imprinted in people’s mind. You don’t contract an architect, you browse a prefab brand website that suits your preferences. Or you might go for the same as your parents did when you grew up.
Another custom which might be unique to Sweden and Finland is to build the house yourselves to various degrees, as it was done traditionally. You order the house from the well-known prefab company that deliver all materials by truck, the prefab company has hired local building contactors to do the time critical assembly of the outer shell to get a weather proof frame. Your contract is with the prefab company and the construction is guaranteed by the prefab company. You don’t have to get to know and trust a local contractor that might go out of business. The rest of the assembly can be left to the home owner to do themselves in their spare time to save on cost, material and building plans delivered and code compliant for ready-to-stanp approval of the authorities. My parents did this 45 years ago while working jobs and raising kids. Doing the DIY route without help from a prefab company is out of reach for most.
Prefab companies also have a very optimized marketing and purchasing process to decide various options, mostly done online. None of my friends or family has hired an architect and then hired a builder. Prefab houses dominate. It is not due to cost primarily I think. A customer will not shop around for the cheapest option, rather go for the safest and most friction-less option for a once in lifetime investment.
The Swedish (Finnish?) prefab industry is worthy of a case study in itself! Wanted to illustrate with one local (small) prefab company from which a friend recently bought a garage/shop that he mostly raised himself. Pay attention to the online 3D CAD construction/customization tool, with on the fly cost calculation.
I suspect that one of the hidden costs of home construction, which is differentially impacting prefab more than manufactured housing, is the complexity of local zoning ordinances.
Manufactured housing is manufactured to occupy lots that are effectively pre-negotiated with local residential zoning authorities. The number of variables left to negotiate and reconfigure to make a particular manufactured home development work is relatively small.
With prefab, the complexity explodes multiplicatively into the same range as regular home construction. Whatever savings there might be labor less the cost of transportation from where the labor is cheaper to where the home is assembled is more than offset by the increased cost of negotiating with local zoning authorities in order to complete the project.
The hypothesis then is that it's the cost of understanding and demonstrating compliance with local ordinances (and, where necessary, negotiating variances) that has prevented prefab from taking off, not the labor and materials that are holding up the projects. It's not a technology problem, it's a human nature problem.
An interesting corollary is that AI may reduce the search and translation costs of compliance with local zoning ordinances. Previously, it would have been completely infeasible to pay humans to try and figure out how a prefab housing plan might be compatible or incompatible with 100s or 1000s of different local zoning requirements. We can do that now. But note that the demonstrating compliance and/or negotiating for variances remains what it has always been.
This is a manifestation of what I have been calling the "synchronization tax" — effectively an extension of Coases's transactions costs to account for the costs of establishing a mutual understanding, which fundamentally requires counterparties to rewrite their descriptions of themselves and the world.
Homes don't require the precision in construction that mechanical equipment does: and the equipment portions of the homes (electrical panels, HVAC equipment, Stoves, etc.) come premanufactured. When there are areas where premanufacturing is helpful, it is done to the portion needed (example: Gluelam beams), not the entirety.
Interesting insight. High tolerances allow for manual labor to substitute well. Plus the cost movement and assembly requires special equipment due to size.
I think this is exactly the mechanism driving the Baumol cost disease for this case. Achieving high precision tolerances is very expensive. It is extremely expensive to do with humans, and while much cheaper to do with scale factory and automation methods, it's still pretty expensive. Building with loose tolerances with a lot of slack is much cheaper, indeed, so much cheaper that investment in capital intensive factory methods is likely to add just as much or more cost as the savings in the wages for the human labor it is intended to replace. Human laborers are ideal and relatively cheap in just this case - when it comes to building items that are cheaper because of loose tolerances and lots of slack. This is because it's actually hard to wring savings out of some hypothetical category of "cheaper because very loose tolerance factory output". To get big- machine-based factories to work at all without dangerously spitting shreds of components all over the place or catching on fire requires an enormous floor of investment in a minimum threshold in high precision production. Consider the example that Musk's engineers at Tesla had tremendous difficulty getting robots to perform certain manufacturing tasks because unless bolts were loaded and grabbed at the precision of a few microns then the equipment would jam or damage product or threaten workers and the assembly would fail. At that time, it was possible to get machines which did it, but the cost was far higher than hiring humans to do it.
The point is, the typical economic gain expected from factory methods is based on an assumption about the production function and the relative substitutability of capital and labor. But the factory methods production function at a higher ratio of capital to labor is constrained by an output that is different because it inherently requires *much* higher quality (precise tolerances, statistical regularity) and even getting started on a production scheme which operates at that higher level of quality requires an initial outlay of capital investment that costs more to carry than just paying human wages. Meanwhile, the final customer *doesn't actually value* that particular type of increase in quality at all, is satiated at a much lower level and doesn't care and is mostly indifferent between the loose-tolerance-high-slack building and the precision manufactured building. So you can't get the factory method product to be much cheaper than the human method product, and you can't get the customer to pay extra for the extra quality he's getting. Thus, you have a robust technologically-and-economically-determined explanation for repeated disappointment for generations and around the world.
We tried to go down the prefab route. In our case there were 0 savings in permitting and in site work, and the overall benefits just weren't that great. and we still needed to find and hire a GC for all the site work.
I've been soo curious about this. I think what could be better than factory built homes is factory built fully structural elements (e.g. 8-10ft tall walls of various lengths) in a partially/mostly-automated factory environment. Local construction teams can simply go to the dwelling-construction-mart for walls, roofs, etc. like they go to a rock/gravel vendor. Thoughts???
SIPS panels are attempting to approach the problem from this angle, but demand is not high enough to allow mass production of the panels in an economically satisfactory manner. Add to that the lack of a highly trained work force capable of assembling said panels into houses that people want.
Build lots of cars, there are roads to put them on. Build lots of houses, which have to be near jobs so the houses can be paid for, and there isn't enough land to put them on...is one more thing.
Housing moves in boom and bust cycles; it's the story of home building. The big guys come and go. Remember US Homes? Exactly. Boom and bust doesn't lend itself to factory production methods. Build the factory, start cranking out homes, and about the time things are going great, it's another bust.
I see, interesting points! SIPS have extra labor to customize them, though... right? The idea I propose is simply studs and nails / screws (maybe a vapor barrier, too) aligned and fastened together by robots - that's it. Although, what are your thoughts on flat-pack structures - standalone residential or standalone commercial?
The problem isn't the framing/wall elements, those go up fast even in regular construction. It's the foundation, plumbing, sewers, flood control, electrical, cable, gas, grading the land, permitting, building codes, land cost, etc . If you had relatively flat, cheap land, easy to dig in, and reasonable wages you might have a chance at cheaper housing.
The simplest fact is the standard single family residence in the forms that Americans demand or will accept is outdated and can't be made "affordable". I've long held a thesis that the American home building industry should be understood as a fashion industry, and is subject to all the sorts of foolishness associated with fashion. If folks want "affordable" housing, different forms have to be accepted.
I use scare quotes on affordable, because there ain't no such thing. "Affordable" is the politician's not so tricky name change from subsidized housing...so as to fool the average American....which is what "affordable" housing in America means. It's not affordable. It's subsidized, and if something has to be subsidized by the government, then by definition it is not affordable to the average person trying to buy a house.
Yes. The past forms cannot be made "affordable", and the new forms I've seen don't fit into the fashion mold demanded by buyers. Municipalities still hold sway on what forms are allowed. Land Use Commissions are almost always populated by old farts happy with the "character of their community" which is a euphemism for "we don't want anything other than what we're accustomed to" and what they're accustomed to (and demand) is adherence to old forms.
Wow...big question. The biggest. I spent 45 years in the gig. I love the idea of flat pack, but the world doesn't. Someday it may. I was recently in the Holon factory at Broad Corp. in Changsha. They're the folks that recently put up the 150 unit high rise in the UAE in...I think it was 96 hours. I think this link will work.
There is a prefab element industry, but mostly for multi-dwelling concrete housing.
For single-family wood frame housing you can go the prefab route but then you need a crane and a crane operator on-site, so your local builder will perhaps not prefer to pay someone out-of-pocket to go that route unless there are specialized companies that cater to this building techniqe and can do it efficiently.
I live in one of the most expensive areas of SoCal. Typical houses cost a fortune (close to a million), and one block over we have cute, clean, affordable single and double wide mobile home (manufactured homes) parks at less than half the cost, even adding on the site rental.
The barrier to affordable housing in this state is political and regulatory. And virtually nobody is talking about this easy solution.
The value of a new house is decided by the value of pre-existing houses in that area. So any cost efficiency gains would not benefit the end consumer. He/she will be willing to pay the same price as for any other house. If new housing were markedly cheaper they would all just sell and move to a newly constructed house. And house values would suffer, which no one wants. So I guess in the end consumer cost will always be decided by the rate of income he/she is willing to spend on housing.
Ita is more a question of where in the supply chain any efficiency gains can be made, and if that particular chain can lock in the profits from that effort. I know nothing of the North American building industry, but am curious if there are any vertically integrated players that could develop next-level production processes to disrupt the industry, or if the industry is to fragmented to allow it?
There are a number of very large builders. The impression I get is that the development end (land purchase, governmental approvals, utility installation, and roads) are a big part of what they do. They will use standard contracts, house plans, and what-not to smooth the process. But after all that, the value in upgrades, to people prepurchasing, dwarfs the benifits of trying to build offsite.
This is great research and wonderful writing , but I was looking forward to an explanation of why the costs never come close
I'm guessing a huge part of it is that land makes up such a vast percent of a house cost.
I'm guessing that the myriad of local and regional zoning makes economies of scale more difficult
But it still seems completely counterintuitive for something that could be craned onto site in one day, can't at least save money from the reduction in time capital is tied up.
volume is too small for real mass-production, so economies of scale (learning curves) don't kick in. so in practice prefab means some guy in a big warehouse doing what some other guy would do on-site.
and as others mentioned due to the heavily cyclical nature of the housing construction industry, coupled with the diminishing returns of adding houses to a specific town/city (it becomes more and more expensive to increase density, even without adding the cost of demolition), plus the other fixed costs (as you mentioned land, as others mentioned permitting, site work, transportation costs and renting cranes, and other specialized heavy machinery) all drastically eat into any possible cost reduction.
again, as someone else mentioned that the home building industry is more like fashion, people really want their own home, pay a lot for perceived quality, and aesthetics (hence a lot of zoning and land use restrictions, HOA-enforced conformism)
also, fundamentally it seems that the amount of matter involved in a building is a few magnitudes more than for a car (and cars have wheels, and there's a huge shared infrastructure for making cars "make sense" - roads, petrol stations, we teach everyone to drive, regulate cars, we spend a lot of money on upkeep of this shared infrastructure, use a lot of precious land for parking, etc.), and to really unlock the efficiency gains from mass-production of buildings the industry would need similar automation and infrastructure (and cost saving profile).
the fact that houses are built and then left where they are for decades means that on-site assembly makes sense, so the parts should be made easily transportable, and assembly must be so fool-proof that then states don't require separate inspections and permitting, and so on. (so it has to be basically self-assembling. or some kind of machine has to do the assembling, that then leaves the site.)
I appreciate this thoughtful comment, and I basically agree with it.
But I do think the tastes of the nation and the type of supply of homes are quite often poorly understood.
People can only buy ( at scale) what's available to buy, at scale.
We're in a self-reinforcing loop, where it looks like prefab won't get the cost benefits of scale, because people don't want them, because they are weird, and thus lending is harder and thus nobody makes them.
We can easily imagine a parallel universe, where some giant corporation essentially produces a brochure of homes, like Sears from before, and people choose from 3 or 4 design themes ( like trims in a car) and from 3 or 4 precise floorplans, with elements open for customization, and we can imagine this fairly limited number of combinations being enough for people, and enough to get scale savings,
But we're not in that universe, Not because it's worse. , But because we have a whole ecosystem that's not designed around us. Permitting, Zoning, Contractors. etc
I'm not saying it's obvious that we will flip to this universe, I'm not saying it's something that would happen soon, But it's a fun universe to think about.
And I don't think we've ever got particularly close, because any prefab solution is either designed to be too cheap or top expensive. for mass market take up.
Well, there's definitely opportunities for scale, standardization, streamlined permitting, better ways to install things safely and reliably (composability, which again depends on standardization), better maintainability (again, standardization, scale), etc.
But the construction industry is very conservative in some sense, and rightfully so. Usually it's clients pushing for new stuff. (Obviously, builders would prefer to build the same thing over and over again!)
> 3 or 4 precise floorplans [...] and enough to get scale savings,
... well, it would definitely help! to some degree this already happens when a big builder builds a new suburb division. (the cookie cutter houses are this, called master-planned communities, yet ... the quality is bad, the precision - as in repeatability of the output - is bad, because it's done by local teams squeezed by the big builder. you can check inspector videos on YouTube.)
.... what's really important to note here is that even though construction is the same everywhere and it's a mature sector, there are so many companies doing it. (and recently there's some concentration) ... this is a symptom which shows how hard it is to find some kind of persistent advantage in the industry. it shows that size doesn't really matter, because apparently even at that size there's not much added efficiency.
It just seems really really obvious that somehow we can do better, yet, it's very hard, because "good enough" is super easy to scale, whereas consistently good is fucking expensive given the constraints (which is your conclusion too)
and, at this point I think we should look at this as a communication and coordination problem. (complex inputs complex outputs, inputs change a lot, outputs change a lot, very hard to know which vendors, builders, architects, agents, were *and are* going to be consistently good over time. so it's like IT, where really hard things also got commodified, like compilers and operating system kernels, and chips, network switches, really scalable things got standardized, and the rest is pure black magic :D)
Brian, thank you for a very interesting article. I feel like you did a very good job explaining the "what" but I still don't understand the "why". Why is homebuilding so resistant to the successes related to cost and quality of other manufacturing processes? Is it the demand for customization? Do we require more options than the~275 models of cars that each have different trims, paints, etc that we can choose from in the US? Is the variability of each site why this is hard? Is it variability land/zoning/entitlements? I feel like each of these things can be tested. Is it something intrinsic to "houses" that is somehow too physically large? Surely its not more complex than advanced technologies we can manufacture.
This article proves without a doubt that factory-built housing is uniquely hard, but I'd love some theories on why this is uniquely hard.
Other commenters have noted two big problems. One is that housing is relatively low precision work. In fact, you want houses designed with a bit of "play" to deal with settling and other stresses. An automobile assembly line worker would have access to a variety of jigs and tools to do the necessary high precision work, but a house assembly line worker would be using the same tools as would be used on site. The other is the lack of scale which presents less opportunity for a learning curve, though a previous Construction Physics article pointed out that even the thousands of Levittown houses had only minimal scale effects.
This captures something I've watched dissolve in real time across fifteen years of owner's rep work.
The factory savings are genuine I've seen them in the quotes. The budget bleeds at the site interface: the moment where the precisely manufactured module meets imperfectly coordinated trades who've never installed that system before and probably never will again.
Construction is one-of-a-kind prototypes assembled by temporary coalitions. The learning curve that makes factories efficient doesn't exist when every project starts from scratch.
Another excellent article. A comment... while the "final assembly" of homes is not industrialized, it seems the supply chain is largely industrialized with common components. If you wanted to compare "industrial" vs "custom" .... wouldn't you want to compare with "custom" components. I suspect that the cost differences are much higher.
Most of the cost of housing is not construction, but the cost of real estate and of rent seeking, arbitrage, and monopolistic behavior. Think private equity and RealPage.
Moving away from piece work can save costs, but most of the reason for the high price of housing is not driven by construction costs.
I think this is the most complete article on this topic I've ever read.
It seems it all boils down to size. If the thing your manufacturing (in this case a house) is bigger than a transportable container its very difficult to compete with. If its smaller its possible but also difficult. What happens when you do a life cycle cost analysis? Do these manufactured homes have the same design lives? Do we have data on that. If so over 60 years can the smaller manufacturered homes even compete?
One question i have is when you say prefabricated do you mean both 2d and 3d?
Good article. I would like to see an analysis that places an economic value on the time savings from using modular construction. E.g. what is the $ value of, say, a 2-month acceleration in start-to-finish turnkey home delivery for the buyer and builder?
One market where manufactured homes might have done well is in rural areas. Our family ranch is in Junction, TX, where there are few builders, electricians, plumbers, etc. Building something here generally means doing it yourself or waiting a long time and paying a lot for one of the few skilled trades people around or using semi-skilled (at best) people who can do parts of a job. Similarly, a friend in Wyoming reports 2 year waits for construction there, due to short building season and lack of labor. Even if not cheaper, the speed of installation would be a great benefit in these sorts of markets.
I would like to know more about how/if CNC and automation factors into the cost of a factory-built home which is then assembled onsite. Surely there could be cost savings as a result of CNC and automated assembly?
Thank you for an interesting article!
Sweden and Finland have huge forestry sectors, with many local prefab industries having evolved near sawmills to capture more of the economic value. With the huge efficiency gains in forestry there has been a surplus of labor and cheap production facilities in these more rural areas.
It is also a mature industry. My parents prefab manufacturer was established 80 years ago. I’d say there are 10 or so strong brands that capture 50% of the market for single-family homes. That is were you start looking if you think of building a home. Different brands have different constructions imprinted in people’s mind. You don’t contract an architect, you browse a prefab brand website that suits your preferences. Or you might go for the same as your parents did when you grew up.
Another custom which might be unique to Sweden and Finland is to build the house yourselves to various degrees, as it was done traditionally. You order the house from the well-known prefab company that deliver all materials by truck, the prefab company has hired local building contactors to do the time critical assembly of the outer shell to get a weather proof frame. Your contract is with the prefab company and the construction is guaranteed by the prefab company. You don’t have to get to know and trust a local contractor that might go out of business. The rest of the assembly can be left to the home owner to do themselves in their spare time to save on cost, material and building plans delivered and code compliant for ready-to-stanp approval of the authorities. My parents did this 45 years ago while working jobs and raising kids. Doing the DIY route without help from a prefab company is out of reach for most.
Prefab companies also have a very optimized marketing and purchasing process to decide various options, mostly done online. None of my friends or family has hired an architect and then hired a builder. Prefab houses dominate. It is not due to cost primarily I think. A customer will not shop around for the cheapest option, rather go for the safest and most friction-less option for a once in lifetime investment.
The Swedish (Finnish?) prefab industry is worthy of a case study in itself! Wanted to illustrate with one local (small) prefab company from which a friend recently bought a garage/shop that he mostly raised himself. Pay attention to the online 3D CAD construction/customization tool, with on the fly cost calculation.
Lövångers Bygg
https://lovangersbygg.se/
I suspect that one of the hidden costs of home construction, which is differentially impacting prefab more than manufactured housing, is the complexity of local zoning ordinances.
Manufactured housing is manufactured to occupy lots that are effectively pre-negotiated with local residential zoning authorities. The number of variables left to negotiate and reconfigure to make a particular manufactured home development work is relatively small.
With prefab, the complexity explodes multiplicatively into the same range as regular home construction. Whatever savings there might be labor less the cost of transportation from where the labor is cheaper to where the home is assembled is more than offset by the increased cost of negotiating with local zoning authorities in order to complete the project.
The hypothesis then is that it's the cost of understanding and demonstrating compliance with local ordinances (and, where necessary, negotiating variances) that has prevented prefab from taking off, not the labor and materials that are holding up the projects. It's not a technology problem, it's a human nature problem.
An interesting corollary is that AI may reduce the search and translation costs of compliance with local zoning ordinances. Previously, it would have been completely infeasible to pay humans to try and figure out how a prefab housing plan might be compatible or incompatible with 100s or 1000s of different local zoning requirements. We can do that now. But note that the demonstrating compliance and/or negotiating for variances remains what it has always been.
This is a manifestation of what I have been calling the "synchronization tax" — effectively an extension of Coases's transactions costs to account for the costs of establishing a mutual understanding, which fundamentally requires counterparties to rewrite their descriptions of themselves and the world.
Homes don't require the precision in construction that mechanical equipment does: and the equipment portions of the homes (electrical panels, HVAC equipment, Stoves, etc.) come premanufactured. When there are areas where premanufacturing is helpful, it is done to the portion needed (example: Gluelam beams), not the entirety.
Interesting insight. High tolerances allow for manual labor to substitute well. Plus the cost movement and assembly requires special equipment due to size.
I think this is exactly the mechanism driving the Baumol cost disease for this case. Achieving high precision tolerances is very expensive. It is extremely expensive to do with humans, and while much cheaper to do with scale factory and automation methods, it's still pretty expensive. Building with loose tolerances with a lot of slack is much cheaper, indeed, so much cheaper that investment in capital intensive factory methods is likely to add just as much or more cost as the savings in the wages for the human labor it is intended to replace. Human laborers are ideal and relatively cheap in just this case - when it comes to building items that are cheaper because of loose tolerances and lots of slack. This is because it's actually hard to wring savings out of some hypothetical category of "cheaper because very loose tolerance factory output". To get big- machine-based factories to work at all without dangerously spitting shreds of components all over the place or catching on fire requires an enormous floor of investment in a minimum threshold in high precision production. Consider the example that Musk's engineers at Tesla had tremendous difficulty getting robots to perform certain manufacturing tasks because unless bolts were loaded and grabbed at the precision of a few microns then the equipment would jam or damage product or threaten workers and the assembly would fail. At that time, it was possible to get machines which did it, but the cost was far higher than hiring humans to do it.
The point is, the typical economic gain expected from factory methods is based on an assumption about the production function and the relative substitutability of capital and labor. But the factory methods production function at a higher ratio of capital to labor is constrained by an output that is different because it inherently requires *much* higher quality (precise tolerances, statistical regularity) and even getting started on a production scheme which operates at that higher level of quality requires an initial outlay of capital investment that costs more to carry than just paying human wages. Meanwhile, the final customer *doesn't actually value* that particular type of increase in quality at all, is satiated at a much lower level and doesn't care and is mostly indifferent between the loose-tolerance-high-slack building and the precision manufactured building. So you can't get the factory method product to be much cheaper than the human method product, and you can't get the customer to pay extra for the extra quality he's getting. Thus, you have a robust technologically-and-economically-determined explanation for repeated disappointment for generations and around the world.
We tried to go down the prefab route. In our case there were 0 savings in permitting and in site work, and the overall benefits just weren't that great. and we still needed to find and hire a GC for all the site work.
I've been soo curious about this. I think what could be better than factory built homes is factory built fully structural elements (e.g. 8-10ft tall walls of various lengths) in a partially/mostly-automated factory environment. Local construction teams can simply go to the dwelling-construction-mart for walls, roofs, etc. like they go to a rock/gravel vendor. Thoughts???
SIPS panels are attempting to approach the problem from this angle, but demand is not high enough to allow mass production of the panels in an economically satisfactory manner. Add to that the lack of a highly trained work force capable of assembling said panels into houses that people want.
Build lots of cars, there are roads to put them on. Build lots of houses, which have to be near jobs so the houses can be paid for, and there isn't enough land to put them on...is one more thing.
Housing moves in boom and bust cycles; it's the story of home building. The big guys come and go. Remember US Homes? Exactly. Boom and bust doesn't lend itself to factory production methods. Build the factory, start cranking out homes, and about the time things are going great, it's another bust.
It's complicated.
I see, interesting points! SIPS have extra labor to customize them, though... right? The idea I propose is simply studs and nails / screws (maybe a vapor barrier, too) aligned and fastened together by robots - that's it. Although, what are your thoughts on flat-pack structures - standalone residential or standalone commercial?
The problem isn't the framing/wall elements, those go up fast even in regular construction. It's the foundation, plumbing, sewers, flood control, electrical, cable, gas, grading the land, permitting, building codes, land cost, etc . If you had relatively flat, cheap land, easy to dig in, and reasonable wages you might have a chance at cheaper housing.
Operative word..."might".
The simplest fact is the standard single family residence in the forms that Americans demand or will accept is outdated and can't be made "affordable". I've long held a thesis that the American home building industry should be understood as a fashion industry, and is subject to all the sorts of foolishness associated with fashion. If folks want "affordable" housing, different forms have to be accepted.
I use scare quotes on affordable, because there ain't no such thing. "Affordable" is the politician's not so tricky name change from subsidized housing...so as to fool the average American....which is what "affordable" housing in America means. It's not affordable. It's subsidized, and if something has to be subsidized by the government, then by definition it is not affordable to the average person trying to buy a house.
Also, what the average American family wanted in 1960 is different than what they want in from 2000 and beyond...
Yes. The past forms cannot be made "affordable", and the new forms I've seen don't fit into the fashion mold demanded by buyers. Municipalities still hold sway on what forms are allowed. Land Use Commissions are almost always populated by old farts happy with the "character of their community" which is a euphemism for "we don't want anything other than what we're accustomed to" and what they're accustomed to (and demand) is adherence to old forms.
Affordable. Affordable for whom?
"better, faster, cheaper - choose 2, can't have all 3" comes to mind... 😕😞
Wow...big question. The biggest. I spent 45 years in the gig. I love the idea of flat pack, but the world doesn't. Someday it may. I was recently in the Holon factory at Broad Corp. in Changsha. They're the folks that recently put up the 150 unit high rise in the UAE in...I think it was 96 hours. I think this link will work.
http://en.broad.com/m/VideoDetail-106-54.aspx
There is a prefab element industry, but mostly for multi-dwelling concrete housing.
For single-family wood frame housing you can go the prefab route but then you need a crane and a crane operator on-site, so your local builder will perhaps not prefer to pay someone out-of-pocket to go that route unless there are specialized companies that cater to this building techniqe and can do it efficiently.
https://youtu.be/CDt0sapaMog?is=aZ8UKxfFENZg5B23
https://youtu.be/sVj1-O7-0Mw?is=usIG5D8pMCzBn49D
IN Australia one of our largest hardware retailers has expanded into this market.
https://www.afr.com/property/commercial/bunnings-invests-75m-in-a-changing-home-build-market-20230904-p5e1u1
Will it work? time will tell!
I live in one of the most expensive areas of SoCal. Typical houses cost a fortune (close to a million), and one block over we have cute, clean, affordable single and double wide mobile home (manufactured homes) parks at less than half the cost, even adding on the site rental.
The barrier to affordable housing in this state is political and regulatory. And virtually nobody is talking about this easy solution.
Super interesting history. But I still didn't understand why (in detail) the cost prefab hasn't dropped.
Materials and permitting as % of build too high?
Assembly on site still too high?
Shipping cost to high?
I'm working on a longer report that will go into detail about this.
The TLDR seems to be that assembly is only 30% of home cost.
Land. Permits. Utility hookup. Financing...
All unaffected by the prefab approach.
So even 50% savings through mfg can only yield 15% in total build deflation.
Need a fully integrated machine for siting/approval AND construction.
This seems like multi family/apartments? Curious why that hasn't really been the case.
I have been told that a big big cost in development is the roads and utility. I guess you could call it site specific costs.
The value of a new house is decided by the value of pre-existing houses in that area. So any cost efficiency gains would not benefit the end consumer. He/she will be willing to pay the same price as for any other house. If new housing were markedly cheaper they would all just sell and move to a newly constructed house. And house values would suffer, which no one wants. So I guess in the end consumer cost will always be decided by the rate of income he/she is willing to spend on housing.
Ita is more a question of where in the supply chain any efficiency gains can be made, and if that particular chain can lock in the profits from that effort. I know nothing of the North American building industry, but am curious if there are any vertically integrated players that could develop next-level production processes to disrupt the industry, or if the industry is to fragmented to allow it?
Value - cost = profit. Your foundational assumption is that markets don't work?
There are a number of very large builders. The impression I get is that the development end (land purchase, governmental approvals, utility installation, and roads) are a big part of what they do. They will use standard contracts, house plans, and what-not to smooth the process. But after all that, the value in upgrades, to people prepurchasing, dwarfs the benifits of trying to build offsite.
I think it's possible that there is a radical rethink of what a house looks like and how it works that is needed to make mass production work.
If you build on site you still need to transport all the material there which must be more costly.
Yes the savings comes from comparing shipping and assembly vs traditional construction
This is great research and wonderful writing , but I was looking forward to an explanation of why the costs never come close
I'm guessing a huge part of it is that land makes up such a vast percent of a house cost.
I'm guessing that the myriad of local and regional zoning makes economies of scale more difficult
But it still seems completely counterintuitive for something that could be craned onto site in one day, can't at least save money from the reduction in time capital is tied up.
But what else explains it
And what can change in the future?
volume is too small for real mass-production, so economies of scale (learning curves) don't kick in. so in practice prefab means some guy in a big warehouse doing what some other guy would do on-site.
and as others mentioned due to the heavily cyclical nature of the housing construction industry, coupled with the diminishing returns of adding houses to a specific town/city (it becomes more and more expensive to increase density, even without adding the cost of demolition), plus the other fixed costs (as you mentioned land, as others mentioned permitting, site work, transportation costs and renting cranes, and other specialized heavy machinery) all drastically eat into any possible cost reduction.
again, as someone else mentioned that the home building industry is more like fashion, people really want their own home, pay a lot for perceived quality, and aesthetics (hence a lot of zoning and land use restrictions, HOA-enforced conformism)
also, fundamentally it seems that the amount of matter involved in a building is a few magnitudes more than for a car (and cars have wheels, and there's a huge shared infrastructure for making cars "make sense" - roads, petrol stations, we teach everyone to drive, regulate cars, we spend a lot of money on upkeep of this shared infrastructure, use a lot of precious land for parking, etc.), and to really unlock the efficiency gains from mass-production of buildings the industry would need similar automation and infrastructure (and cost saving profile).
the fact that houses are built and then left where they are for decades means that on-site assembly makes sense, so the parts should be made easily transportable, and assembly must be so fool-proof that then states don't require separate inspections and permitting, and so on. (so it has to be basically self-assembling. or some kind of machine has to do the assembling, that then leaves the site.)
... see also https://gulfconstructiononline.com/Article/388450/Abu_Dhabi%E2%80%99s_16-storey_Earth_Tower_installed__in_96_hours ( http://en.broad.com/m/VideoDetail-106-54.aspx )
I appreciate this thoughtful comment, and I basically agree with it.
But I do think the tastes of the nation and the type of supply of homes are quite often poorly understood.
People can only buy ( at scale) what's available to buy, at scale.
We're in a self-reinforcing loop, where it looks like prefab won't get the cost benefits of scale, because people don't want them, because they are weird, and thus lending is harder and thus nobody makes them.
We can easily imagine a parallel universe, where some giant corporation essentially produces a brochure of homes, like Sears from before, and people choose from 3 or 4 design themes ( like trims in a car) and from 3 or 4 precise floorplans, with elements open for customization, and we can imagine this fairly limited number of combinations being enough for people, and enough to get scale savings,
But we're not in that universe, Not because it's worse. , But because we have a whole ecosystem that's not designed around us. Permitting, Zoning, Contractors. etc
I'm not saying it's obvious that we will flip to this universe, I'm not saying it's something that would happen soon, But it's a fun universe to think about.
And I don't think we've ever got particularly close, because any prefab solution is either designed to be too cheap or top expensive. for mass market take up.
Well, there's definitely opportunities for scale, standardization, streamlined permitting, better ways to install things safely and reliably (composability, which again depends on standardization), better maintainability (again, standardization, scale), etc.
But the construction industry is very conservative in some sense, and rightfully so. Usually it's clients pushing for new stuff. (Obviously, builders would prefer to build the same thing over and over again!)
https://www.construction-physics.com/p/what-construction-innovation-uptake
> 3 or 4 precise floorplans [...] and enough to get scale savings,
... well, it would definitely help! to some degree this already happens when a big builder builds a new suburb division. (the cookie cutter houses are this, called master-planned communities, yet ... the quality is bad, the precision - as in repeatability of the output - is bad, because it's done by local teams squeezed by the big builder. you can check inspector videos on YouTube.)
.... what's really important to note here is that even though construction is the same everywhere and it's a mature sector, there are so many companies doing it. (and recently there's some concentration) ... this is a symptom which shows how hard it is to find some kind of persistent advantage in the industry. it shows that size doesn't really matter, because apparently even at that size there's not much added efficiency.
It just seems really really obvious that somehow we can do better, yet, it's very hard, because "good enough" is super easy to scale, whereas consistently good is fucking expensive given the constraints (which is your conclusion too)
and, at this point I think we should look at this as a communication and coordination problem. (complex inputs complex outputs, inputs change a lot, outputs change a lot, very hard to know which vendors, builders, architects, agents, were *and are* going to be consistently good over time. so it's like IT, where really hard things also got commodified, like compilers and operating system kernels, and chips, network switches, really scalable things got standardized, and the rest is pure black magic :D)
https://www.builderonline.com/builder-100/builder-100-list/2025/
https://eyeonhousing.org/2025/07/top-10-builder-market-share-across-metros/
Brian, thank you for a very interesting article. I feel like you did a very good job explaining the "what" but I still don't understand the "why". Why is homebuilding so resistant to the successes related to cost and quality of other manufacturing processes? Is it the demand for customization? Do we require more options than the~275 models of cars that each have different trims, paints, etc that we can choose from in the US? Is the variability of each site why this is hard? Is it variability land/zoning/entitlements? I feel like each of these things can be tested. Is it something intrinsic to "houses" that is somehow too physically large? Surely its not more complex than advanced technologies we can manufacture.
This article proves without a doubt that factory-built housing is uniquely hard, but I'd love some theories on why this is uniquely hard.
Other commenters have noted two big problems. One is that housing is relatively low precision work. In fact, you want houses designed with a bit of "play" to deal with settling and other stresses. An automobile assembly line worker would have access to a variety of jigs and tools to do the necessary high precision work, but a house assembly line worker would be using the same tools as would be used on site. The other is the lack of scale which presents less opportunity for a learning curve, though a previous Construction Physics article pointed out that even the thousands of Levittown houses had only minimal scale effects.
This captures something I've watched dissolve in real time across fifteen years of owner's rep work.
The factory savings are genuine I've seen them in the quotes. The budget bleeds at the site interface: the moment where the precisely manufactured module meets imperfectly coordinated trades who've never installed that system before and probably never will again.
Construction is one-of-a-kind prototypes assembled by temporary coalitions. The learning curve that makes factories efficient doesn't exist when every project starts from scratch.
Another excellent article. A comment... while the "final assembly" of homes is not industrialized, it seems the supply chain is largely industrialized with common components. If you wanted to compare "industrial" vs "custom" .... wouldn't you want to compare with "custom" components. I suspect that the cost differences are much higher.
Most of the cost of housing is not construction, but the cost of real estate and of rent seeking, arbitrage, and monopolistic behavior. Think private equity and RealPage.
Moving away from piece work can save costs, but most of the reason for the high price of housing is not driven by construction costs.
I think this is the most complete article on this topic I've ever read.
It seems it all boils down to size. If the thing your manufacturing (in this case a house) is bigger than a transportable container its very difficult to compete with. If its smaller its possible but also difficult. What happens when you do a life cycle cost analysis? Do these manufactured homes have the same design lives? Do we have data on that. If so over 60 years can the smaller manufacturered homes even compete?
One question i have is when you say prefabricated do you mean both 2d and 3d?
Good article. I would like to see an analysis that places an economic value on the time savings from using modular construction. E.g. what is the $ value of, say, a 2-month acceleration in start-to-finish turnkey home delivery for the buyer and builder?
One market where manufactured homes might have done well is in rural areas. Our family ranch is in Junction, TX, where there are few builders, electricians, plumbers, etc. Building something here generally means doing it yourself or waiting a long time and paying a lot for one of the few skilled trades people around or using semi-skilled (at best) people who can do parts of a job. Similarly, a friend in Wyoming reports 2 year waits for construction there, due to short building season and lack of labor. Even if not cheaper, the speed of installation would be a great benefit in these sorts of markets.
Yes just buying something like this would seem to speed up the process.
https://lovangersbygg.se/sortiment/stall/
Yes this seems to be the selling point. Less variety (good and bad). More predictable build time.
Very interesting article - thank you!
I would like to know more about how/if CNC and automation factors into the cost of a factory-built home which is then assembled onsite. Surely there could be cost savings as a result of CNC and automated assembly?