Not long ago we looked at construction productivity trends for the US and for countries around the world. We found that in the U.S., and in most other large, wealthy countries, construction productivity is stagnant or declining. Unlike manufacturing and agriculture, or the economy overall, which generally show improving productivity over time, in the field of construction we find that productivity tends to at best stay constant, and at worst decline over time.
this sounds like we should look at what's different in japan, sweden and switzerland, where the construction costs are at least rising slower than the CPI
They're not rising slower than CPI. They're rising more slowly than the US's consumer price index, but they're rising at or above the consumer price index in those respective countries. Construction costs have historically been close to flat in Japan because inflation has been close to flat, but costs are still rising faster than inflation there.
Thank you for this. I'd be interested in an attempt to break down where the costs are coming from. Like, I expect the cost of labor is going up over time as it is in other industries (which is a good thing!). Given little to no automation in construction we should expect costs to rise right?
What about worldwide indices of materials costs? Surely these goods should fall in price since they are manufactured goods? And what about equipment costs?
One clarification: is the price of land included in these indices in some way? If so, increasing urbanization and rising land prices would drive cost increases.
In Japan, the same arithmetic shows up on the manufacturing side. The experienced line left the floor through retirement and restructuring, and was replaced by foreign labor and new hires on short contracts. The small judgments at the line, the errors caught before they ran — that work did not transfer. Productivity stopped rising. The cost column held its number; the workforce behind the number had changed. What stays constant in the index is rarely what stays constant in the shop.
I am a big fan of your book. But I have gone down a bit of a rabbit hole on the construction of cpis. And the issue of hedonic quality adjustments means that comparing construction costs to cpis, especially in the USA makes very little sense. And comparing across countries makes even less sense. Riksbanken in Sweden, for example has estimated that the Swedish cpi is 0,5%-points lower annually than the rest of Europe, solely because they do larger hedonic adjustments for it etc. many countries also have huge problems with chaining effects in clothing nd other categories where things regularly goes on sale before new model/collection comes in. Quality adjustments for buildings are much harder to do.
So many questions. Why hasn't the cost of construction been less than the overall level of inflation, given it's a process amenable to learning; it's "making" a house, or an apartment building, and generally things that are "made" can benefit over the long term from advances in technology and technique, basically from learning and scaling? Or is it that "most" tasks - laying some flooring, say - see no ability to go much faster than back in the day? Drywall is drywall?
Yet even that question takes me for a trip. Most people would be happy, I suspect, if construction costs that follow inflation meant prices that also merely followed inflation, but it seems that in housing, all the headlines are about prices running amok? Implying construction costs FAR exceeding inflation.
We now live in a building in DC built around 2020, 26 apartments, half one bedroom, half two-bedroom, a few variations, concrete, steel/no wood, drywall, and all the modern additions like fire and sprinkler systems, elevator, sensors for all this like the Starship Discovery, it seems, and big glass windows. It appears to have "cost" about $5-7M total (I was curious and did a little digging) and the builder, once all was said and done, sold at about a 3X multiple of this. Yet I saw where putting in a single elevator in a subway in NYC was running $72M - or 10 of these buildings! In such a world, do "averages" stop having much meaning? (OK, maybe mixing "costs" and "budgets"?)
Similarly, disconnects. I recall my first property in 1991, a townhouse about 1 year old at the time, priced at about 2.5X my gross annual salary. Today, that property costs a person with the same job position (Federal employee in Florida) 5X gross annual salary. "Inflation" and "averages" begin a story, but would it not make sense to see construction costs in relation to salaries - to follow the plot - these things we build are homes for people after all.
How about looking at "what the market will bear". Housing is a large part of the wealth of a country, and wealth recently has been inflating more than wages. Government statistics are defined to exclude wealth (by definition the value is the value) but if you step back there is clearly more wealth chasing a fairly constant pile of assets, and the increasing concentration of wealth is visible in statistics. In the USA the inflation of wealth has run around 7% per year for the past few decades. And the housing market runs on trading up plus the minting of loans based on assets, clearly the market for houses will bear what the wealth will pay.
Your graphs show costs growing slower than inflation, if you consider wealth as its market. In other words, they are getting more efficient, but the money keeps spiralling in.
That's a good one. As in the thesis of Capital in the 21st Century, if we assume houses and such are tracked as "capital," not as the "economy." Inflation adjustments would seem linked to the economy, not markets, which implies that the basis for analyzing construction as part of the economy starts with a mismatch.
Maybe looking at construction input or output cost is too narrow of a viewpoint. How about profits? Share price is a surrogate for margin and ROI. As an example, DR Horton has a nearly 10,000% share price increase since 1992. It is one component of the XHB homebuilders index.Pulte has a 70,000% share price increase over the somewhat longer term! Obviously some of this is upselling to those that can afford a new home, but most is likely market power of the sector.
Contrast this with St Louis FRED data for construction wages, which roughly doubled in their time series data from 2006 to present (not inflation adjusted):
It seems some sectors and economic classes benefit from increasing construction costs while others do not. The story of neoliberalism and finance capitalism, no?
Maybe the only way to substantially reduce housing costs (not so much construction costs) is to reduce the amounts spent on land, labor, and financing, which is a subject I discuss rather amateurishly in my book, chapter 3, note vi: https://www.amazon.com/dp/B00U0C9HKW
Quite insightful analysis, especially the distinction between productivity metrics and actual construction cost behavior over long time periods.
One thing I kept thinking about while reading this: how much of the persistent cost inflation in construction comes not just from labor/material inefficiencies, but from coordination uncertainty in increasingly complex projects?
In existing buildings especially, teams are often working with fragmented documentation, undocumented modifications, and assumptions that only get validated during construction. In practice, a significant portion of “cost growth” seems tied to discovering reality too late in the process.
Do you think modern digital workflows like Scan to BIM, point cloud-based existing condition modeling, and coordinated BIM environments have the potential to meaningfully change long-term construction cost trends or do they mainly redistribute costs earlier in the project lifecycle rather than reduce them overall?
The quality adjustment problem you describe points to something worth naming: construction cost indexes measure what goes into a building, not what it delivers over its operational life.
Structural standards and energy performance can at least be proxied. But whether a building actually supports the work happening inside it (over 20 or 30 years of occupancy) has never been part of the measurement framework.
Indoor environmental monitoring is starting to generate data that could change that. Whether it eventually feeds into how we assess building value is an open question. The cost-per-square-foot lens will keep missing it either way.
I wonder whether some of the increased costs of building are related to higher standard and customer expectations. If land is making my house cost significantly more maybe I spend more per sqft because I barely notice the difference.
This is a useful framing because it separates “technical improvement” from “cost improvement”.
One thing I’ve seen from the construction automation side is that a new process rarely lowers cost simply because the tool is more advanced. The cost only starts to move when the whole operating system around it changes: design constraints, permitting logic, QA, training, logistics, project sequencing, inspection, liability allocation, and repeatability across jobs.
That may be one reason construction absorbs technical improvements without seeing the same cost declines as manufacturing. The innovation often enters as a specialised task improvement, but the project delivery system remains largely unchanged.
For automation, I think the key diligence question is not “can the robot do the task?” but “does the technology simplify the delivery system enough to change the cost structure?”
Nice work. Costs relative to each country, fx adusjted over time would be more meaningful to me, rather than 1 country relative to their own respective baseline.
this sounds like we should look at what's different in japan, sweden and switzerland, where the construction costs are at least rising slower than the CPI
They're not rising slower than CPI. They're rising more slowly than the US's consumer price index, but they're rising at or above the consumer price index in those respective countries. Construction costs have historically been close to flat in Japan because inflation has been close to flat, but costs are still rising faster than inflation there.
You summarized the article's labrynthian complexity succinctly for me !
Thank you for this. I'd be interested in an attempt to break down where the costs are coming from. Like, I expect the cost of labor is going up over time as it is in other industries (which is a good thing!). Given little to no automation in construction we should expect costs to rise right?
What about worldwide indices of materials costs? Surely these goods should fall in price since they are manufactured goods? And what about equipment costs?
One clarification: is the price of land included in these indices in some way? If so, increasing urbanization and rising land prices would drive cost increases.
In Japan, the same arithmetic shows up on the manufacturing side. The experienced line left the floor through retirement and restructuring, and was replaced by foreign labor and new hires on short contracts. The small judgments at the line, the errors caught before they ran — that work did not transfer. Productivity stopped rising. The cost column held its number; the workforce behind the number had changed. What stays constant in the index is rarely what stays constant in the shop.
Might I ask about the elephant in the room- how is China doing?
I am a big fan of your book. But I have gone down a bit of a rabbit hole on the construction of cpis. And the issue of hedonic quality adjustments means that comparing construction costs to cpis, especially in the USA makes very little sense. And comparing across countries makes even less sense. Riksbanken in Sweden, for example has estimated that the Swedish cpi is 0,5%-points lower annually than the rest of Europe, solely because they do larger hedonic adjustments for it etc. many countries also have huge problems with chaining effects in clothing nd other categories where things regularly goes on sale before new model/collection comes in. Quality adjustments for buildings are much harder to do.
So many questions. Why hasn't the cost of construction been less than the overall level of inflation, given it's a process amenable to learning; it's "making" a house, or an apartment building, and generally things that are "made" can benefit over the long term from advances in technology and technique, basically from learning and scaling? Or is it that "most" tasks - laying some flooring, say - see no ability to go much faster than back in the day? Drywall is drywall?
Yet even that question takes me for a trip. Most people would be happy, I suspect, if construction costs that follow inflation meant prices that also merely followed inflation, but it seems that in housing, all the headlines are about prices running amok? Implying construction costs FAR exceeding inflation.
We now live in a building in DC built around 2020, 26 apartments, half one bedroom, half two-bedroom, a few variations, concrete, steel/no wood, drywall, and all the modern additions like fire and sprinkler systems, elevator, sensors for all this like the Starship Discovery, it seems, and big glass windows. It appears to have "cost" about $5-7M total (I was curious and did a little digging) and the builder, once all was said and done, sold at about a 3X multiple of this. Yet I saw where putting in a single elevator in a subway in NYC was running $72M - or 10 of these buildings! In such a world, do "averages" stop having much meaning? (OK, maybe mixing "costs" and "budgets"?)
Similarly, disconnects. I recall my first property in 1991, a townhouse about 1 year old at the time, priced at about 2.5X my gross annual salary. Today, that property costs a person with the same job position (Federal employee in Florida) 5X gross annual salary. "Inflation" and "averages" begin a story, but would it not make sense to see construction costs in relation to salaries - to follow the plot - these things we build are homes for people after all.
How about looking at "what the market will bear". Housing is a large part of the wealth of a country, and wealth recently has been inflating more than wages. Government statistics are defined to exclude wealth (by definition the value is the value) but if you step back there is clearly more wealth chasing a fairly constant pile of assets, and the increasing concentration of wealth is visible in statistics. In the USA the inflation of wealth has run around 7% per year for the past few decades. And the housing market runs on trading up plus the minting of loans based on assets, clearly the market for houses will bear what the wealth will pay.
Your graphs show costs growing slower than inflation, if you consider wealth as its market. In other words, they are getting more efficient, but the money keeps spiralling in.
That's a good one. As in the thesis of Capital in the 21st Century, if we assume houses and such are tracked as "capital," not as the "economy." Inflation adjustments would seem linked to the economy, not markets, which implies that the basis for analyzing construction as part of the economy starts with a mismatch.
Maybe looking at construction input or output cost is too narrow of a viewpoint. How about profits? Share price is a surrogate for margin and ROI. As an example, DR Horton has a nearly 10,000% share price increase since 1992. It is one component of the XHB homebuilders index.Pulte has a 70,000% share price increase over the somewhat longer term! Obviously some of this is upselling to those that can afford a new home, but most is likely market power of the sector.
Contrast this with St Louis FRED data for construction wages, which roughly doubled in their time series data from 2006 to present (not inflation adjusted):
https://fred.stlouisfed.org/series/CES2000000003
It seems some sectors and economic classes benefit from increasing construction costs while others do not. The story of neoliberalism and finance capitalism, no?
Maybe the only way to substantially reduce housing costs (not so much construction costs) is to reduce the amounts spent on land, labor, and financing, which is a subject I discuss rather amateurishly in my book, chapter 3, note vi: https://www.amazon.com/dp/B00U0C9HKW
I'm sure Brian could do a much better job!
Quite insightful analysis, especially the distinction between productivity metrics and actual construction cost behavior over long time periods.
One thing I kept thinking about while reading this: how much of the persistent cost inflation in construction comes not just from labor/material inefficiencies, but from coordination uncertainty in increasingly complex projects?
In existing buildings especially, teams are often working with fragmented documentation, undocumented modifications, and assumptions that only get validated during construction. In practice, a significant portion of “cost growth” seems tied to discovering reality too late in the process.
Do you think modern digital workflows like Scan to BIM, point cloud-based existing condition modeling, and coordinated BIM environments have the potential to meaningfully change long-term construction cost trends or do they mainly redistribute costs earlier in the project lifecycle rather than reduce them overall?
Would genuinely love your perspective on that.
The quality adjustment problem you describe points to something worth naming: construction cost indexes measure what goes into a building, not what it delivers over its operational life.
Structural standards and energy performance can at least be proxied. But whether a building actually supports the work happening inside it (over 20 or 30 years of occupancy) has never been part of the measurement framework.
Indoor environmental monitoring is starting to generate data that could change that. Whether it eventually feeds into how we assess building value is an open question. The cost-per-square-foot lens will keep missing it either way.
I wonder whether some of the increased costs of building are related to higher standard and customer expectations. If land is making my house cost significantly more maybe I spend more per sqft because I barely notice the difference.
What is the theory for this relationship of construction cost increases and general inflation?
This is a useful framing because it separates “technical improvement” from “cost improvement”.
One thing I’ve seen from the construction automation side is that a new process rarely lowers cost simply because the tool is more advanced. The cost only starts to move when the whole operating system around it changes: design constraints, permitting logic, QA, training, logistics, project sequencing, inspection, liability allocation, and repeatability across jobs.
That may be one reason construction absorbs technical improvements without seeing the same cost declines as manufacturing. The innovation often enters as a specialised task improvement, but the project delivery system remains largely unchanged.
For automation, I think the key diligence question is not “can the robot do the task?” but “does the technology simplify the delivery system enough to change the cost structure?”
Nice work. Costs relative to each country, fx adusjted over time would be more meaningful to me, rather than 1 country relative to their own respective baseline.