Updated: Feb 26
Commentary by Craig Webb, president, Webb Analytics
Go to a meeting about modular construction, mass timber, and other off-site components manufacturing and odds are one of the speakers will show the chart above. I've been to events where four consecutive speakers showcased it in their presentations. Some of these speakers work at companies like Katerra, the Silicon Valley startup that's spending huge sums on mass timber and other projects in a bid to revolutionize how Americans build. Others organize investments, develop properties, run architectural firms, or promote themselves as futurists.
I have used the chart in my own speeches. Now I'm beginning to question using it. Perhaps more important, I'm wondering whether investors should make big bets based on it.
The chart appears on page 23 of Reinventing Construction: A Route to Higher Productivity, from the McKinsey Global Institute. Published in February 2017 by a unit of the world-renowned consultancy McKinsey & Co., the report asserts that construction needs to dramatically improve its productivity rate. To make its case, McKinsey uses the chart to argue that the U.S. agriculture industry became 16 times more productive between 1945 and 2010 and manufacturing improved its productivity 8.6 times over the same period, but, construction is just 1.1 times better--and since the 1980s, construction productivity has actually declined. This particular chart is so important to McKinsey's argument that it's referenced in the opening paragraph of the report's preface.
I believe speakers embraced this chart because it backs their gut feeling that construction needs to change. Typically, it is preceded by slides showing, say, a picture of a construction project in the 19th century, and then a disturbingly similar scene from today. After that setup, presenting the McKinsey data nails home the point that builders are way behind their peers.
McKinsey's report is must reading for anyone interested in how to build better. But Greg Brooks--like me, a former editor of ProSales--smelled something fishy in that particular, oft-cited chart. How could McKinsey's productivity numbers for construction be flat, he asks, when we now have prehung doors, factory-built cabinets, panel sheathing, drywall instead of plaster, power tools, and a myriad of other improvements?
It turns out that some of Brooks' answer lies in a technical appendix tucked into the back of McKinsey's 166-page report. In that appendix, McKinsey notes a big caveat to its numbers: When the construction process gets better because of something a building product manufacturer did, the gain shows up in manufacturing's productivity index, not construction's. Only when prefabrication is performed by a construction company do productivity gains accrue to the construction sector.
That raises a question: Does a maker of trusses, joists, and panels count as a manufacturer or a construction firm? Identifying them as manufacturers means the construction industry doesn't reap any of the speed and accuracy benefits that stem from using components built off-site. This classification problem is sure to get tougher as builders embrace "cassettes," in which open-web joists will come to the job site with HVAC and plumbing equipment already installed. Should they be considered a manufactured product, or as part of construction? (The photo shown here displays a MiTek-designed cassette)
McKinsey also noted that the statistics it relied on for its country-by-country comparisons don't include migrant or undocumented workers. "Their exclusion artificially inflates official productivity statistics in countries with non-negligible levels of informal labor," McKinsey said, thus presumably making countries like the U.S. look worse in comparison.
The challenge measuring construction productivity that McKinsey admits to in its back pages is a front-burner issue for economists such as the University of Chicago's Chad Syverson and Austan Goolsby. It's common in other industries to measure productivity by dividing the number of products produced by the number of workers producing the products. An auto factory with 100 workers that turned out 1,000 cars a week in 2007 but 1,500 cars per week in 2017 clearly is more productive. But when you try to do that by measuring single-family permit issuances per construction workers, the resulting graph (shown here) reveals no clear trend beyond what the Great Recession and subsequent recovery can do to the productivity line.
Goolsbee and Syverson have joined forces with the U.S. Labor Department's Bureau of Labor Statistics to examine what should go into a useful construction productivity report. Their initial results? "Yes, the official productivity numbers for the sector are horrible," Syverson wrote in an email to me. "The magnitude of the horribleness is definitely tied to what price index one uses to deflate revenues to output, but even the more generous scenario implies slow growth as opposed to outright decline."
"It does seem that the sector is lagging," Syverson continued. "It is not manufacturing, that’s for sure. We’re still trying to figure out why."
So.... How close to the truth about construction productivity is McKinsey's chart? Probably not as precise or as bad as the consulting company indicates. Then again, when your major points of comparison improved by 16 and 8.6 times, respectively, you could show much better numbers--a 4x improvement vs. the stated 1.1x, for instance--and still be regarded as pitiful. And what McKinsey created still would have value.
All this brings me to another chart and a new story. The chart below comes from a major tract builder that estimated the cost of building one of its homes using stick-built methods vs. paying an off-site manufacturer to create panels.
For this builder, the savings it reaped on raw materials, installation, and dumpster fees were just about matched by what it paid the panel maker. Going off-site only was marginally better.
The story is about a U.S.-funded factory in southernmost Mexico that my father was responsible for maintaining back in the mid-1970s. The plant covered a couple of football fields, yet all the cement for it was produced using a 12-cubic-foot cement mixer and several hundred local workers with wheelbarrows.
One day, a leader for the workers came to the foreman and announced the workers were going on strike.
"What do you want?" the foreman asked.
"Higher wages," the strike's leader replied.
"What are you paid now?" the foreman then asked. The strike leader's answer in pesos converted to something like 30 cents a day. This was a time in which Mexico was devaluing its currency regularly, so the U.S. dollars being used to pay for the project had built up close to double its previous buying power.
"How much more do you want?"
"35 cents a day."
"Sure. Get back to work."
Could this project have been far more productive had the foreman bought in a truck-sized cement mixer and used far fewer workers? Yes, but labor was so cheap and plentiful that a less productive technique won out. And so it is today. Again and again in construction, it's the chart with dollar signs on it that's going to prevail.