Abundant incrementalism
The fastest path to transformative supply
This is the last post in a series on Incremental Development. Previous posts in this series:
When I talk about incremental development with pro-housing folks, especially those who live and work in the most supply constrained metros, they tend to nod along and “agree” with the basic idea. But the most common pushback I get is something like, “we need to do more than incrementally increase housing supply, we need a transformative increase in supply!”
I think that question masks an important misunderstanding. When we talk about incremental development, we are talking about the scale of the individual lot, not the scale city or metro. Housing markets function at the scale of the city and metropolitan area — not individual properties, but the sum of all of them. And unlocking incremental development at the scale of the lot is in fact the most transformative thing we can do at the scale of the city or metropolitan area, because it impacts every lot.
To close out my thoughts on incrementalism I want to walk you through a few simple thought experiments that illustrate this point, and hopefully help you embrace a new mindset. I was initially going to call this something like “ambitious” or “expansive” incrementalism, but instead I’m going to shamelessly follow the zeitgeist and call this mindset Abundant Incrementalism.
Different paths to housing supply
I’ll use Denver as the model for my argument today. Denver is a popular and growing city experiencing supply constraint. According to the Denver Regional Council of Governments, the area is currently 52,000 units short of balanced supply, and needs to add 511,000 units by 2050 to maintain balanced supply in the face of expected growth. For this example, let’s take those numbers at face value1: to deliver 511k units in 25 years, then we need to add 20,440 units per year. That’s the goal, now how do we get there?
First, as a baseline, let’s consider what it would take to deliver all that housing via suburban expansion. If we take a typical suburban density of 5,000 people per square mile2, and the Colorado average household size of around 2.5 people per household, that means “typical” suburbia contains about 2,000 homes per square mile. So to deliver 511k homes, we’d need to develop about 256 square miles of new land3, or about 10 square miles of new development per year. The land is available, but horizontal development at that scale isn’t politically popular, and it requires a ton of capital for new infrastructure that often fails to pay for itself, so relying solely on horizontal expansion isn’t fiscally prudent either.
Second, let’s think about filling in downtown. There’s about 100 surface parking lots in downtown Denver. While these significantly vary in size, for napkin math, if we assumed that we could build large-scale buildings with an average of 200 apartment units per project, we could net 20k units total. But that’s only solving for one year of supply… and realistically we couldn’t do it in a year, it would be hard to deliver those 20k units in the entire 25 year window we’re looking at.
Now, there’s plenty of underdeveloped land outside of downtown, so we could find other locations for apartments. But remember, that aggressive building plan only solved for one year of supply. And 200 units is a relatively large apartment building. Where would be put the next 100 large apartment buildings that we’d need to build every year? It’s physically possible but historically subject to tremendous political backlash.4 I’m also skeptical that we could find enough apartment developers and investors to fund and construct multifamily buildings at that pace, even if there was no political obstacle.
Finally let’s consider what incremental development could deliver.
Denver itself (not the metro area) has around 20k blocks5, and a typical block has 20-25 lots on it. If policy change made it easy to add one unit to any existing lot (via a backyard cottage, accessory apartment, duplex conversion, etc.), and we then added one unit per block, then we’d add 20k units per year.6 We could do that for 20-25 years without repeating lots, gradually “thickening up” the city.7 And this could continue indefinitely. In 25 years, many of those early projects would be fully depreciated and ready for further redevelopment — imagine an old house divided into a duplex this year, and 25 years from now torn down and rebuilt as a new 4-plex.
Note that I just looked at the city of Denver, not the metro area. Denver today is about 720k people out of a metro population of 2.9M, so about 25% of the metro area. I have no idea how many blocks there are across the metro, but if this scales with population then 3/4 of the lots in the metro are outside the city. If those incrementally redeveloped at the same pace, we could add 80k units per year via incremental development, knocking out the metro area’s supply shortage in year one and reaching out 25 year goals in less than 7 years. If that pace of development was sustained, the Denver metro would more than double, growing by 5 million people in the next 25 years. That would unlock truly abundant supply, enough that the limit on additional housing would be demand, not permission.8
The power of the denominator
A short aside to emphasize why incremental development is so powerful. In established urban areas, already served by infrastructure, the most fundamental, irreducible cost of urban development is the cost of the land.9
When you subdivide a large lot in the city and build two houses on it, you’re reducing the land cost by 50%. To get the next 50% you’d have to cut the land in half again, going to four units, then eight, sixteen, etc. So when we think about driving down cost, the biggest and easiest cost reduction is by that first halving.
This effect is less powerful when building new subdivisions, because you’re generally starting from a very low land cost in the first place (or else it wouldn’t pencil) and you are paying a high cost to install the necessary infrastructure for the first time. And the effect is less powerful on large-scale development because larger buildings eventually require more expensive construction techniques and internal infrastructure (parking garages, elevators, sprinkler systems etc.) that end up dominating the land cost.
This is all just to say that, in America, where the majority of homes are single family houses, accessory units and duplex conversions are the lowest hanging fruit and the largest-scale opportunity to increase housing supply.
Transformative change
To close, I want to emphasize this series is not an argument against any form of development.
In a timely coincidence,
wrote a great post today explaining that the government shouldn’t be in the business of telling people where to live. I fully agree. We should be building housing where there is demand for housing, which means that we should allow large-scale projects in our center cities and incremental development throughout our established neighborhoods and high quality horizontal expansion on the edges. We should allow the next increment by right in every neighborhood.What I want you to take away from today’s post is that, in every American metro area, the biggest unlock on the table is to allow the specific thing our cities forbid today: letting existing homeowners do small projects that add one unit.
And this isn’t something for us to grudgingly accept as some compromise, taking what we can get. This is the reform we should be the most excited about; a transformative change that unlocks abundant housing supply.
Of course nobody has a crystal ball and these numbers could be wrong, but we don’t need a precise forecast so much as we need a reasonable placeholder to check our thinking against.
5,000 people per square mile is being generous. That’s closer to mature, older suburbs like Englewood and more than double average of current fast-growing suburbs like Parker and Commerce City.
For context, this is about double the size of Denver city limits, but only a fraction of the metro area.
It’s also not clear to me if how much demand there is for apartments as a housing type… I think that way more people would choose to live in apartments than do today if those apartments were both pleasant and affordable, but it’s pretty hard to do affordable new construction, and also hard to guess how filtering would play out over a generation of rapid multifamily development. So, who knows.
Denver's city block count isn't officially documented, but based on its grid system, we can estimate. Denver's total area is about 154.7 square miles, with about 16 blocks per mile east-west and 8 blocks per mile north-south in most areas. This suggests approximately 128 blocks per square mile, resulting in around 19,800 city blocks across Denver. This is a very coarse estimate, and I think this is on the low side (many parts of town have smaller blocks than this), so I rounded up.
Note that this pace of incremental development, limited just to the city of Denver itself, is almost exactly the amount and pace of supply that DRCOG says we need to deliver for the metro area. This is an uncanny coincidence — I swear, I did the math on what incremental supply could unlock first, and then went back to look for a forecast of what the Denver metro needed.
And, living in a neighborhood that’s undergoing this exact kind of change today, I feel confident in stating that this is perceived as “gradual.” While the addition of one backyard cottage or duplex conversion each year on each block does is noticeable, that kind of steady gradual change is nowhere near as noteworthy as delivering 100 large apartment buildings across town every year. That matters in as much as change that’s perceived to be “big” or “sudden” tends to produce more political backlash than smaller, gradual changes.
Case in point, I doubt there’s enough demand for Denver to add 5M residents in the next 25 years… I think there might be an early wave of growth, and then abundant supply would just look like a boring, stable housing market where prices mostly keep pace with inflation regardless of how much the city grew (or didn’t grow) that year.
Land cost can be artificially high due to speculation, etc., and there are things we could try to counter that. However, development always makes the surrounding land more valuable, so even if we were able to counteract all speculation in the market, the value of land in urban areas will naturally be high in high demand places.
The more experience I have with megaprojects, the more jaundiced I am about them, and the more zealous I become for distributed action, the way you describe here. Let the flowers bloom!
This is great Andrew. Doing the basic math is so important here… and so rarely done.