I apprenticed with the Carpenters' Union just after the establishment of OSHA, in 1971. Most of my work was off the ground, forming large concrete structures and building scaffolding for other crafts to work safely off the ground. That first year OSHA implemented regs requiring us to be tied off to safety lines at all times when we were more than thirty feet off the ground. We resisted because it felt like complying slowed us down too much. My attitude changed when I returned to work eighteen months after shattering my right leg in a twenty-five foot fall. While I had learned patience, it was clear that safety came at the cost of productivity.

Expand full comment

I think your overall conclusion is correct, but it is worth analogizing the struggles in construction safety with the struggles in construction automation, because they come from the same source.

It is easier to identify and mitigate safety issues in manufacturing processes for the same reason that it's possible to automate them: we have near-absolute control over the built environment in which they take place. We can add safety features to machinery, including shielding rotary assemblies and driveshafts from human contact, providing climate control, incorporating hand shields and reducing access around sharp objects, etc. We can train workers on processes that will be the same, day in and day out, down to the literal motion of their limbs around a piece of machinery.

These things often do mildly impact productivity relative to the baseline of no safety features or training (sewing machines are an emblematic case), but because they're often introduced simultaneously to other machinery and process improvements and because they reduce disruption, downtime, and turnover, they are only a small drag on productivity.

Moreover, safety equipment, by its nature, is most cumbersome when being moved, installed, or uninstalled. Thus it mainly impacts maintenance in factories, not production. But safety equipment on construction sites is *always* being moved and reinstalled as part of routine operations, so the impact is worse because the nature of site work makes safety measures more routinely invasive.

In addition, because we already struggle to implement any kind of productivity-enhancing automation, safety improvements are arriving on their own, unaccompanied by the sort of broader process and equipment improvements which would mask their impact. When industrial sewing machines began introducing needle guards, for example, they were simultaneously pioneering reversible functions and computer-aided stitching, so the slight slowdown that most experienced seamstresses will admit to seeing with a needle guard was hidden in the wash.

Fall protection for steel erectors, meanwhile, came alone, with no such productivity advances to mitigate it.

Overall I agree that safety practices are a drag on productivity rather than the main determining factor, but they are more challenging for the construction sector for the same reason that automation is more challenging.

Expand full comment

Not exactly tied to productivity, enough to bring up a website that fascinates me and that I link with the slightest excuse: Reclamation's list of fatalities for Hoover dam


This list is sorted into pages by year, then by type of fatal accident, and finally sorted by date, giving a by-name list and cause for deaths.

I had toured the dam when I was a kid, and had heard the statistics about the number of people killed (I think the 112 number on that page). However, until I was working in the construction industry, I did not realize just how abso-posi-loutely batsh*t bonkers this list is. It is one of the most astonishing documents to ponder when you have a twenty-first century view about the acceptability of construction casualties.

The one I always point at is the list for 1933: https://www.usbr.gov/lc/hooverdam/history/essays/fat1933.html

I really urge everybody who is familiar with how construction accidents to be handled on a modern jobsite to contemplate this project when deaths are considered chronologically:

Jan 1: Fred Palmer, killed by fall

Jan 10: Gus Enberg, killed by explosion

Jan 11: Howard Cornelius, killed by explosion*

Feb 1: Vernon Blair, killed by fall

Feb 7: M. Kaighn, killed by falling material

Feb 7: J. Powers, struck by truck/equipment

Mar 12: Tom Markey, killed by fall

Mar 16: William Koontz, struck by truck/equipment

For the first quarter of that year, they were killing a guy every two weeks on average on this jobsite. Using today's standards, by mid-February the US Government would fire this contractor. This is one of my favorite(?) examples of "The past really is a different country."

* Same location as Enberg the day before, but it's not clear if this was one event, that killed them both, or a <i>separate accident of the same type a day later</i> but as this list shows, the latter is absolutely plausible.

Expand full comment

When you say "Similarly, while the puddling process for making iron required back-breaking physical labor, it wasn’t especially dangerous – in 1910 there were no recorded fatalities out of a total of 1300 puddlers working in the US.", that accounts for deaths by *accident*, but it was hard on your body. Wikipedia notes "The strenuous labour, heat and fumes caused puddlers to have a very short life expectancy, with most dying in their 30s."

Expand full comment

Wow great i have grown up in the UK and Australian Construction and i concur with your article, with increased competitiveness in our sector and the rise in interest rates, over the last 12 months 1260 construction companies have gone into administration in our Australian Construction industry great article id love to discuss this with you more.

Expand full comment

doesn't the rise and dominance of Platform Framing have a lot to do with light wood construction jobsite safety? where it's rare to be more than 8' away from a supporting floor? Also prefabricated roof trusses plus large cranes for installing them Versus the pre-Platform days of post-and-beam, where every board, even high roof members, is hand lifted and hand placed and hand nailed.

Expand full comment

Table 2 seems like it only applies to Single Family Homes. As described in the article, these are much safer to construct than high rises. Are OSHA and safety regulations in general a bigger factor in bigger buildings? This research paper: https://www.researchgate.net/publication/331281055_The_cost_of_implementing_OHSMS_regulation_on_high-rise_building_projects says it is 2.01%-3.70% of costs. That is for Indonesia, but Bing says it is comparable to the U.S.:

"According to a Business Round Table report (Construction Industry Institute, 1988), the cost of an effective construction safety and health program in the USA is approximately 2.5% of direct labor costs. Successful safety programs have been developed by many construction companies and have shown remarkable results."

It's much cheaper to build high in most parts of the world than in the U.S., especially in America's largest city, New York: https://buildingtheskyline.org/skyscraper-height-iv/. This helps to explain why most of the highest high rises now go up outside the U.S., but safety costs are only a small part of that labor factor.

Expand full comment

“It’s hard to escape the conclusion that it’s at least part of the answer. But I suspect it’s only a part, perhaps not a particularly large one. For one, as we saw previously, homebuilders don’t seem to think the burdens of OSHA and other safety regulations are all that high”

Is it possible that home builders rate safety compliance costs low, because they do not comply with OSHA as much as the rest of the construction industry? I don’t think I have ever seen a single family home being constructed without obvious safety violations. Usually no tie off points, no mid rails or kick plates on scaffolding, etc

Expand full comment