Fragmentation in the building industry

Fragmentation in the building industry
Art by Wu Guanzhong

An excellent framing of the fragmentation in the building industry can be found in Dana A Sheffer’s 2010 dissertation on the Diffusion of Energy Saving Technologies in the Building Industry. (Also see related work here by Daniel Hall.) Sheffer’s piece inspired my own thinking, as written here:


The decentralized nature of the building industry must be understood in order to make impactful improvements to how infrastructure is built and maintained. Much of this stems from the organization of collaboration across the industry, globally.

Modular vs. Integral linkages

Herbert Simon discusses organization as nested hierarchic structures; the most successful are “nearly decomposable” into modular components.

Modularity allows for “black box” replacement as a strategy, increasing competition through substitutes and incremental innovation. Think about changing your refrigerator for a more efficient one. Installing a “smarter” elevator. (Or even hiring for roles in an organization.) Change happens subsystem by subsystem. This retains stability. But what do you do when you want to change the system architecture as a whole?

Before we can answer that, we need to specify the distinction between centralized and decentralized modular structures. Centralized modular structures can be found in the aerospace and automobile industries. Lead firms direct alignment with the relevant power.

On the flip side, decentralized modular structures are the name of the game in the construction industry. This structure is self organizing and coordinates through mechanisms like standards (e.g. building codes) or status (e.g. programs like “LEED”).

While modularity allows for changing individual subsystems, it’s very hard to change the interfaces of these subsystems. Interface change is more successful for organizations that have a centralized structure: the lead firm can align direction. However, for decentralized systems, like for the buildings industry, “disparate firms need to self synchronize.” This is not easy.

Origins of fragmentation

Historically, building construction began as master builder, an integrated system. Over time, the industry fragmented vertically into “design” then “build” then “operate,” with a new team at each stage. This vertical fragmentation also began to extend horizontally. Complexity led to specialization. For each stage, multiple disciplines (craft trades) now partake in separate fashion. Just think of the combinatorics!

Now, extend that fragmentation even further. Due to the bid / tender process, a NEW project team is created for EACH new construction or retrofit building. This creates longitudinal fragmentation: new teams repeated over and over.

As people, we celebrate working with diverse teams, but what are the switching costs for project delivery?

Finally: let’s look at the modularity of the building components themselves. Let’s take air conditioning as an example. As shown below, this is longitudinally fragmented, too. A new vendor for each stage of the process. Not only does this staging happen in the design and install, but in the operations too! Data is siloed, and end value accrues to multiple entities.

Why does this matter?

First, it’s imperative to think about your choice of strategy when introducing a new technology or framework to the construction / buildings industry. New innovations should choose their path wisely– whether they are modular or interface-dependent.

Second, it’s clear that this structure lacks the transfer mechanisms for operations and maintenance. Each modular entity only cares about the completion of their stage; they’re not incentivized for the ongoing life of the whole system. To use an analogy from economist Tyler Cowen– who is not speaking about the industry per se but rather about specialization within our society– “the incentive today is to build a brick, not to build the building.” And certainly not to operate that building for the betterment of occupants.

So what’s the next question?

How might decentralized structures better foster ongoing operations and maintenance? The answer is likely in standards and status.