It’s often hard to find valuable academic research on construction - the number of useful papers seems low, beyond just what would be implied by Sturgeon’s Law. It largely seems written for an audience of other academics, and tends to be fairly disconnected from the nuts and bolts of actually putting a building together.
So I was pleased when I stumbled across the paper “The Construction Industry as a Loosely Coupled System”, by Anna Dubois and Lars-Erik Gadde. Buried behind the academic jargon is an interesting model of the construction industry that helps explain why it is the way it is.
They start with an interesting reframing of the construction efficiency question - though construction is often criticized for failing to be quote-unquote productive enough, we should be impressed that buildings manage to get successfully built at all.
The Complexity of the Building Process
Buildings are complex artifacts: a typical single family home has somewhere in the neighborhood of 3,000 parts, and an apartment building might have 10 to 100 times as many depending on the size. For comparison, a typical car has somewhere around 30,000 parts, and a Boeing 737 has around 600,000.
These parts are put together in a specific way, in a specific order, by dozens of different workers. Assembly tasks are highly interdependent, and each step of the process depends on the previous steps being completed successfully - install the insulation wrong, and you’ll delay the installation of the drywall, which pushes back trim, which pushes back painting, etc. Construction in some ways faces a uniquely difficult assembly task - the site-built nature of the product means it’s hard to parallelize component production (I can’t get a jump on the electrical by starting it before the framing is done), and the trade-specific nature of the tasks means there’s little labor flexibility (if your framer doesn’t show up, you can’t take half the HVAC crew to start cutting studs so the project stays on schedule).
Building production is also characterized by a great deal of uncertainty and factors outside the builder’s control. The physical environment is often difficult to predict (whoops, we got seven days of rain, whoops, we found a huge boulder when we were excavating), as is the cultural and regulatory environment (whoops, the zoning board is making us remove 20 units, whoops, the plan reviewer is saying we need to add more brick). On top of this, the building design process results in much less information compared to other types of production - the drawings used to put up the building don’t come with a step-by-step set of instructions, or a complete list of parts.
A building is built by many different firms working together, each one handling a small aspect of the overall process. The plumbing contractor installs the plumbing, the mechanical contractor installs the HVAC, etc. Most complex production processes follow a similar model - the Boeing 777 has engines made by Pratt & Whitney, tail rudders made by ASTA, and electronics made by Honeywell. But what sets construction apart is the nature of these firm relationships.
Production of other complex artifacts tends to require close relationships between the firms doing the work. Firms develop strategies for working with each other, spend the effort to adapt to each other’s workflows and processes, and create close partnerships that can last for years or decades. Boeing’s suppliers would have engineers stationed on-site at Boeing HQ to help resolve coordination and manufacturing problems. Lear recently built a factory in Flint specifically to manufacture seats for GM. Prior to the pandemic, Apple was reportedly spending $150 million per year on flights to China, where most of their manufacturing and suppliers are located.
In construction, on the other hand, relationships between firms are much weaker and shorter term. Project teams change frequently from project to project, and being selected for a project team is often largely a function of being the low bidder. As a result of these short term, transactional relationships, firms typically don’t spend time or effort adapting to each other's processes or developing close working relationships. If anything, relationships between construction firms tend toward adversarial, as fixed-fee bidding forces everyone to carefully manage their scope.
Given these difficult production conditions, how do buildings manage to successfully get built?
A Community of Practice
The answer lies in the way the industry is structured - construction practices are in large part governed by a set of standards and processes that exist outside any one firm.
For one, buildings are generally constructed out of fairly standardized components - different contractors will all use the same type of dimensional lumber, the same type of roof shingles, the same standard steel sections, etc. Often this standardization takes the form of specifications published by trade industry bodies or associations, such as the APA’s plywood performance standards, or the ASTM standard governing masonry block properties. Other times, suppliers will converge on nearly interchangeable products (nearly all hollowcore plank suppliers offer 8” x 48” sections, for instance). Outside of vertically integrated design builders (common in things like parking garages or metal buildings), most builders will use easily available off the shelf components - it’s rare for project-specific or builder-specific components to be used.
These standards also exist in the form of building codes, which give a uniform set of (often very specific) rules that buildings must follow, limiting the extent that one building can differ from another. Building codes differ from state to state, but nearly every state adopts some version of the International Building Code and then makes minor adjustments to it.
But beyond documented standards, the industry is also strongly shaped by what Dubois and Gadde refer to as a “community of practice” - a combination of informal norms, widely accepted techniques, craft skills, and unspoken assumptions that together create a common set of expectations on a project. When a team comes together for a project, even if they’ve never worked together before, there will already be a shared set of expectations as to how information will be represented on drawings, the deliverables the project will require, who will have responsibility for what, how the different systems will interact, and all the other items that might otherwise require time, effort, and expense to coordinate.
This community of practice is reinforced by contractual requirements and state laws which require contractors and designers to conform to “standard practice” or meet the “standard of care” - in essence, requiring you to do what a reasonable professional would do.
Tradition and culture in construction is thus load bearing - “the way we’ve always done it” is the answer to how we’re able to do it all. This method of organization stems from construction’s history as a collection of tasks performed by skilled trades, workers hired for their expertise (often acquired under an apprenticeship), and given wide leeway to accomplish their tasks.
Community of Practice and Advanced Framing
This sort of industry organization has benefits and drawbacks. Light framed wood construction provides a good example.
For a conventional wood framed house, a set of construction documents can be surprisingly sparse without impacting the builder’s ability to build it successfully. The drawings don’t need to show where each individual stud goes, or how construction needs to be sequenced, or even the sorts of headers that need to go over doors and windows. The designer can rely on the International Residential Code (which gives prescriptive header sizes) to provide much of the necessary information, and the skills and experience of the framing crew to fill in most of the gaps. This allows a building to be successfully designed and built with a comparatively small design budget.
The flip side of this is that you’ll face a lot of difficulty if you deviate from standard practice. An example of this is Advanced framing, or Optimum Value Engineering (OVE). Advanced framing allows a house to be built with less material and higher energy efficiency by removing “extra” wood members that aren’t needed to support the house (studs at 24 inches instead of 16 inches, 2 stud corners instead of 3, single top plates instead of double, etc.)
It seems like a trivial optimization, but it is shockingly difficult to get builders to adopt advanced framing. The details and assembly methods are different from conventional wood construction, it differs from what the building code prescribes (though the code allows it), and it’s not what framers have generally been trained to do - it exists outside the community of practice. Most advice for how to get a house built with advanced framing is “find a builder familiar with green building techniques” i.e.: find a builder in a different community of practice.
Community of Practice and Innovation
This industry structure adds another potential difficulty for construction innovation - conventional construction projects get a large amount of the work “for free”, in the form of existing expertise, standards, and expectations. If I want to get a house built, I can rely on the builder and the subcontractors to figure out most of what needs to be done without needing to be told.
But this goes out the window if I want to use some new, innovative building system unlike anything else on the market - now I DO need to tell them exactly what to do, and exactly how to do it. Even if I expect it to be better and cheaper than conventional methods, I can expect it to incur extra design costs (since I can’t rely on the builder or the code to fill in the gaps), and extra training costs. I’ll also probably need to accept an initial lower level of productivity and an increased number of mistakes, as the workers learn how to use the new system.
What’s more, these costs won’t be one-time - I’ll be paying them over and over again every time I get a new project crew or use a new contractor.
This puts conventional construction in a sort of local optimum - most new methods of building, even potentially cost-saving ones, will likely face an extended period of being more expensive and more difficult than conventional methods.
So, to sum up:
Buildings are complex objects, but are built differently than other complex objects: construction firms tend to have weaker, more transactional relationships than firms in other industries.
The construction industry overcomes this by relying on a set of shared standards, assumptions, and norms that exists outside of any one firm, and act as a sort of ambient coordination between firms with otherwise weak relationships.
This “community of practice” is what makes it possible to construct buildings at all, but it also makes it hard to innovate - any new method or technique will likely face an initial period of being worse than conventional methods.
These posts will always remain free, but if you find this work valuable, I encourage you to become a paid subscriber. As a paid subscriber, you’ll help support this work and also gain access to a members-only slack channel.
Construction Physics is produced in partnership with the Institute for Progress, a Washington, DC-based think tank. You can learn more about their work by visiting their website.
You can also contact me on Twitter, LinkedIn, or by email: firstname.lastname@example.org
Another good article Brian. I have noted that "the way we always do it" is stronger in single family residential (reaching a peak in production housing) than institutional buildings. There is a direct correlation to the detail of the design documents and also design fees. The design effort and documentation for a production home is very sparse, and yet they bang them out by the thousands. A courthouse today will have hundreds of full size drawings and thousands of pages of specs.
Also, the 3000 parts count in a house is very low. If you count each shingle, brick tie, and brick, each electric face plate, and floor tile (each part handled by a human installer), you get closer to 30 000 than 3000. There are over a thousand asphalt shingles in a modest house. Toyota says that their cars have over 30 000 parts but they count fasteners "down to the smallest screws" (and they often show up on parts diagrams). If you count fasteners, then 30 000 parts in a single family home is easily exceeded. I cant see how a larger building like a federal courthouse would not have 600 000.
Complexity and Lean thinking is defined by the time it takes to complete. I have heard of homes in Phoenix built in 30 days, my friend does complete bank renovations in 2 weeks working 24/7. In China a highrise floor every 48 hours. Any studies on applying lean to construction?