Look around the room you’re sitting in. In the room are a bunch of products. Those products will have been made in one of two ways.
One way companies make products is to handle things from start to finish: design the products, manufacture them and sell them.
In the kitchen in which I’m sitting, I can see a clay jug from some Irish pottery company. The potter sourced the clay and made the jug and sold the jug. I don’t own a Tesla, but if I did, there would be a car outside the house that had been both designed and manufactured by Tesla.
The car I actually own, by contrast, was designed by one European company and assembled by a couple of dozen global companies, working in collaboration. Back in the kitchen, I see a TV and some cookbooks. The TV wasn’t manufactured by Sony, and the cookbooks weren’t printed by their publishers.
Today I want to talk about the logic of doing it yourself versus collaborating with others. I want to talk about it, not in the context of Tesla and kitchenware, but in that of state investment in infrastructure projects.
Good enough?
The business theorist Clay Christensen gave us the framework for thinking about this . He said that there’s a tradeoff between a product’s performance and its cost to produce.
The cheapest way to produce something is to get a load of commodity components and bolt them together. For many products, that’s just what you want. My Sony TV works perfectly well, and it was really cheap. My cookbooks are nice. I don’t need them to be any nicer. I’m not willing to pay extra for additional niceness when it comes to cookbooks.
But there are other products where you really do want them to be good, and you’re willing to pay extra for the quality. The iPhone is an imperfect example because it does outsource manufacturing. But Apple nonetheless is the most integrated phone maker. It designs both the hardware and the software, by contrast to say Samsung, which uses off-the-peg software from Google.
This is the reason why Apple makes approximately all the profit in the smartphone category. When it comes to phones, many people want the best performance possible, and they’re willing to pay for it.
It’s a similar story with my kitchen jug. The whole point of the jug is that it’s really nice to look at and feels special. I don’t want my potter in Letterfrack to email their jug design to some contract manufacturer in Jingdezhen, China. When it comes to jugs, I care more about them being good than being cheap.
Tesla is an interesting one because it’s the only car company to design and build its cars. The rest of the industry takes the approach that consumers are price-conscious and that it makes more sense to save money by outsourcing than to maximise performance by producing everything in-house. Tesla’s big bet is that electric cars are fundamentally different, and that people are going to be willing to pay for better performance.
Teslas’ performance is better because fewer things prevent Tesla engineers from fixing performance bottlenecks. If a Tesla engineer wants to make it so the car seat can be adjusted on the touch pad, they can go ahead and do that. A Volkswagen car seat might be manufactured by Bosch, and the seat might have to be able to interface with a dozen car companies to whom Bosch sells seats, so the fancy touch screen car seat integration doesn’t happen in Volkswagen cars.
On the other hand, Volkswagen probably sources its car seats more cheaply than Tesla, because Bosch specialises in car seats and manufactures them at scale. Swings and roundabouts.
Construction
That’s one part of it: the tradeoff between performance and price. It’s cheaper to bolt together commodity components. Performance is better when everything is designed from scratch.
In the construction business, the most common setup is for the designer and the builder to be separate entities. You don’t buy a kitchen extension from a kitchen extension company. You pay a designer and pay a builder.
One reason for this is because a kitchen is a kitchen. You don’t need ultra-high-performance floors and windows, designed specially for your house. Commodity floors and windows do the job fine. When it comes to windows, you care more about price than performance.
There is another consideration though, beyond price and performance. It’s relevant to the construction industry. The other consideration I’ll call ignorance. People who pay for builders usually don’t know anything about building. The buyer often can’t tell if the work has been done right — whether the flashing on the roof has been installed correctly.
To the extent that buyers are ignorant, it’s another argument for separating design from building. The architect’s job is not just to design the thing, but to oversee its construction on behalf of the buyer. They become the buyer’s agent. They make sure the flashing is on right. They verify the work.
For most construction jobs, this setup works fine. The builders bolt together commodity materials, verified by the architect. This has two big benefits for the buyer. The first is that they get an agent who looks out for their interests. The second is that the project gets built cheaply. Builders and architects operate on very narrow margins and, from a buyer’s perspective, that’s great.
There is, however, a fourth and final thing to consider when deciding whether it’s wise to outsource part of a construction project. That thing is complexity. The more complex a project, the harder it is to bolt together. Beyond a certain level of complexity, having the design team at arm’s length from the builders is more trouble than it’s worth.
One problem is that in an arm’s-length relationship, the builders have different incentives to the designers. The designer/client’s incentive is for the job to be done as well as possible; the builders’ incentive is for the job to be done as cheaply as possible. In a small project, these conflicts are relatively simple. In a large one, the conflicts are highly complex. So the relationship between builder and designer/client gets adversarial.
Then there are the practical problems of coordinating very complex jobs, in which there are hundreds of thousands of steps, which all must be done in sequence. The bigger the job, the harder it is to coordinate these steps. This coordination challenge is made even harder if the builder has separate interests to those of the designer/client.
Here’s how Brian Potter described the challenge of complex projects in his newsletter Construction Physics:
“… as building complexity increases, the costs of verifying construction has been done correctly rise SLOWER than the potential savings from coordinating – the costs of verifying rise linearly with the number of building elements, whereas the savings from coordinating rise with the square of number of building elements (since each new element needs to be coordinated with every existing element).”
State projects
When it comes to capital investment by the State, the two models for delivery are design-build and design-bid-build. In design-build, the same entity is responsible for both design and construction. In design-bid-build, the design function and the construction function are separate.
In Irish capital spending, though there are some design-build projects, we lean heavily on design-bid-build. The National Children’s Hospital was a design-bid-build project. And it suffered from all the pathologies I described above in terms of coordination costs and adversarial relationships between builders and their clients.
For cookie-cooker projects, like say schools, it might be the case that design-bid-build is the way to go. But Ireland is contemplating much bigger, more complex projects, Metrolink being one.
An implication of all this is that for complex projects, the state is better off integrating the design and construction stage. The downside is that it doesn’t have competitive bidding for the construction phase and pays contractors higher margins. The upside is that it can adequately coordinate the project.
The following chart shows the benefits of each delivery method in the context of a complex metro line:
And here’s the same chart for a secondary school:
Mega-projects get a bad rap. Part of the reason would appear to be a reliance on design-bid-build. As we’ve seen previously, design-build metro lines get built at 10-20 per cent the cost of design-bid-build ones. Why do countries persist with the flawed design-bid-build model for complex capital projects? One reason is that design-build puts an onus on the state to take on hard, complex work. Instead of paying an agent to run the job, like a homeowner paying an architect, the state would be expected to run the job themselves. It would be expected to have the staffing, the experience, and the risk appetite to get the job done.
Getting to a point where the Irish state has the chops to handle complex projects will be hard. But if we want complex infrastructure projects, it’ll surely be easier in the long run.