We spend a lot of time around here looking at construction costs, and whenever we do the same story emerges: construction never gets cheaper. Construction costs tend to rise at or above the level of overall inflation. A square foot of building costs as much or more to build today as it did in the 1950s.
I notice your discussion and charts about the rate of price changes relative to CPI over time are a little bit misleading. If I understand correctly, you are comparing the rate of change in CPI over each time period to the rate of change of each process' price during the same period, setting up %changeCPI -%changeProcessPrice to get the over/under percentages. However, you then say that most of the changes are about zero, with histogram buckets of -.3 to 1.7%. That is actually a huge range considering the data you are looking at, because the CPI change is in single digit percentages.
From 1954-1985 the CPI grew ~8.8%, and from 1985-2022 ~5.2% (according to https://www.rateinflation.com/ and my quick math.) Therefore a changing price 1% greater than CPI between 54 and 85 is growing at a rate 11% faster than CPI; 1% above CPI 85-22 is a 19% greater rate. In other words, a bucket spanning roughly 2% absolute range around zero should definitely hold the vast majority of your data, because it would be shocking to have points far outside there.
To look at it another way, at 5% inflation prices double roughly every 14 years, but at 6% inflation they double every 11.6 years, 17% less time from an addition of 1% to the rate (adding 20% of the base rate).
Perhaps this points to an increase in the total number of tasks being inserted into housing.
Let's say a very simple and humble SFH in the 50's was put up - they might not even have put insulation in the walls in a temperate environment or be required by a permitting office to do so, or only installed 1 outlet per room.
If the number of tasks to build a modern home has twice the number of steps or tasks that alone could explain increases in costs, even if each individual task had gotten significantly more cost effective over time.
This makes me curious about where productivity increases/falling materials costs are most transparent *to the purchaser.* Why should the contractor charge me so much less as one element gets cheaper if all the constructors are pretty confident I don't know enough to say, "Hey, that's a nail gun, I want a discount over the old, hammer rate!" Plus, prices are sticky, so a small improvement in the cost side seems more likely to offer a cushion than to be a reason to slash prices.
Very interesting! Here are two hypotheses about your discrepancy.
1) Actual increases in construction labor hourly rates are lower than captured by the data sources you used. The obvious culprit here would be the use of undocumented day laborers who might significantly bring down the overall average costs.
2) There's "shared overhead" of some of the setup that's not captured in individual task estimates. This would represent a real productivity gain that's not necessarily captured. For instance, let's take hand driven nails vs. pneumatic nail workflows. Imagine that it would take a while to set up your compressor at the beginning and end of the day. If all you're doing is hanging joists, the set up and take down time is "costed" to hanging joists. That would make sense. But in reality, construction teams may leave their compressor in place at the end of the day, so they can use it for the next day or for future tasks unrelated to joist hanging. In that case, the "setup time" generally costed to a task, would be avoided and the task would be on average more efficient than reflected in a time studies.
In any case, I'd love if you could get to the bottom of the discrepancy with someone at RS Means. They probably wouldn't share their whole methodology, but they might have some ideas.
Regarding the discrepancy in productivity. I am not sure this is particularly surprising. The US, though less trade-dependent than many other Westerns economies, is open enough that wages will be set by the maginal labour productivity gain. If this is higher than in the domestic/sheltered sector, you will get a situation where the actual productivity (installations per day) will vary from the labour productivity (installations per wage, or value added per employee)
I think it would add more to compare construction costs before and after the creation of the Federal Reserve. There's something called the Skyscraper Index (also called the Skyscraper Curse) in which economic downturns coincide with the construction of tall Skyscrapers. There are also Cantillon Effects and the Austrian Business Cycle theory which I think may play a role in why prices increase and decrease seemingly at random while the overall costs track with inflation, especially considering inflation is tied to the Fed's policies.
I’m not sure I’m clear on the first chart’s methodology.
Are you saying that, for each year, you took the rate of cost increase for each item and subtracted the CPI-derived rate of inflation, then took the average of each year’s values? Because that number is basically meaningless. An average rate over time is not the same as a cumulative compounding effect. If an investment loses 40% this year and gains 40% next, the average rate of change is zero but the cumulative one is -16%.
If you’re saying you took the cumulative increase over those periods and subtracted the change in CPI over the same time, then the data has value as presented.
Productivity, is something that works on a learning curve, which requires a focused drive. GM and Ford drive Productivity. Chemical plants drive Productivity. There is capital Productivity, ie $ investment/outputUnit goes down. Then there is Variable Cost Productivity. Lean, Quality and value analyis of low value steps are reduced. Materials or parts integration. DFMA on the assembly process.
A. The electronics industry reigns supreme. Trillions of things made.
B. Automotive is awesome- millions lf things made
C. Aerospace is weak.. 1 to 1000s made
D. Medical Device is horrible at learning curve
E. Agriculture is stagnant.
F. Construction - I would imagine if you looked at high capital intensive plants, ie Oil, chemical, maybe auto plants (or model changeovers) vs retail housing, maybe it's different.
But basically, in Construction, who and what entities are in charge of driving Productivity ?
I notice your discussion and charts about the rate of price changes relative to CPI over time are a little bit misleading. If I understand correctly, you are comparing the rate of change in CPI over each time period to the rate of change of each process' price during the same period, setting up %changeCPI -%changeProcessPrice to get the over/under percentages. However, you then say that most of the changes are about zero, with histogram buckets of -.3 to 1.7%. That is actually a huge range considering the data you are looking at, because the CPI change is in single digit percentages.
From 1954-1985 the CPI grew ~8.8%, and from 1985-2022 ~5.2% (according to https://www.rateinflation.com/ and my quick math.) Therefore a changing price 1% greater than CPI between 54 and 85 is growing at a rate 11% faster than CPI; 1% above CPI 85-22 is a 19% greater rate. In other words, a bucket spanning roughly 2% absolute range around zero should definitely hold the vast majority of your data, because it would be shocking to have points far outside there.
To look at it another way, at 5% inflation prices double roughly every 14 years, but at 6% inflation they double every 11.6 years, 17% less time from an addition of 1% to the rate (adding 20% of the base rate).
Another consideration:
Perhaps this points to an increase in the total number of tasks being inserted into housing.
Let's say a very simple and humble SFH in the 50's was put up - they might not even have put insulation in the walls in a temperate environment or be required by a permitting office to do so, or only installed 1 outlet per room.
If the number of tasks to build a modern home has twice the number of steps or tasks that alone could explain increases in costs, even if each individual task had gotten significantly more cost effective over time.
This makes me curious about where productivity increases/falling materials costs are most transparent *to the purchaser.* Why should the contractor charge me so much less as one element gets cheaper if all the constructors are pretty confident I don't know enough to say, "Hey, that's a nail gun, I want a discount over the old, hammer rate!" Plus, prices are sticky, so a small improvement in the cost side seems more likely to offer a cushion than to be a reason to slash prices.
Overall construction in down in quality and durability.
Very interesting! Here are two hypotheses about your discrepancy.
1) Actual increases in construction labor hourly rates are lower than captured by the data sources you used. The obvious culprit here would be the use of undocumented day laborers who might significantly bring down the overall average costs.
2) There's "shared overhead" of some of the setup that's not captured in individual task estimates. This would represent a real productivity gain that's not necessarily captured. For instance, let's take hand driven nails vs. pneumatic nail workflows. Imagine that it would take a while to set up your compressor at the beginning and end of the day. If all you're doing is hanging joists, the set up and take down time is "costed" to hanging joists. That would make sense. But in reality, construction teams may leave their compressor in place at the end of the day, so they can use it for the next day or for future tasks unrelated to joist hanging. In that case, the "setup time" generally costed to a task, would be avoided and the task would be on average more efficient than reflected in a time studies.
In any case, I'd love if you could get to the bottom of the discrepancy with someone at RS Means. They probably wouldn't share their whole methodology, but they might have some ideas.
Regarding the discrepancy in productivity. I am not sure this is particularly surprising. The US, though less trade-dependent than many other Westerns economies, is open enough that wages will be set by the maginal labour productivity gain. If this is higher than in the domestic/sheltered sector, you will get a situation where the actual productivity (installations per day) will vary from the labour productivity (installations per wage, or value added per employee)
I think it would add more to compare construction costs before and after the creation of the Federal Reserve. There's something called the Skyscraper Index (also called the Skyscraper Curse) in which economic downturns coincide with the construction of tall Skyscrapers. There are also Cantillon Effects and the Austrian Business Cycle theory which I think may play a role in why prices increase and decrease seemingly at random while the overall costs track with inflation, especially considering inflation is tied to the Fed's policies.
Maybe another way to think about it would be to ask "why are construction costs such a good proxy for the overall inflation rate?"
I’m not sure I’m clear on the first chart’s methodology.
Are you saying that, for each year, you took the rate of cost increase for each item and subtracted the CPI-derived rate of inflation, then took the average of each year’s values? Because that number is basically meaningless. An average rate over time is not the same as a cumulative compounding effect. If an investment loses 40% this year and gains 40% next, the average rate of change is zero but the cumulative one is -16%.
If you’re saying you took the cumulative increase over those periods and subtracted the change in CPI over the same time, then the data has value as presented.
Fascinating analysis!
Productivity, is something that works on a learning curve, which requires a focused drive. GM and Ford drive Productivity. Chemical plants drive Productivity. There is capital Productivity, ie $ investment/outputUnit goes down. Then there is Variable Cost Productivity. Lean, Quality and value analyis of low value steps are reduced. Materials or parts integration. DFMA on the assembly process.
A. The electronics industry reigns supreme. Trillions of things made.
B. Automotive is awesome- millions lf things made
C. Aerospace is weak.. 1 to 1000s made
D. Medical Device is horrible at learning curve
E. Agriculture is stagnant.
F. Construction - I would imagine if you looked at high capital intensive plants, ie Oil, chemical, maybe auto plants (or model changeovers) vs retail housing, maybe it's different.
But basically, in Construction, who and what entities are in charge of driving Productivity ?