Updating my priors on missing middle

I haven’t been that vocal about it, but I have tended to be of the opinion that missing middle housing - legalized duplexes and triplexes, accessory dwelling units (ADUs), etc. - are mostly a symbolic victory. Just generally getting the local busy-bodies out of the way for very minor changes to each neighborhood’s structural environment probably isn’t going to make a huge dent in housing affordability, but it is a sign of a polity doing the right things. A city that passes a pro-ADU law is to a YIMBY like a local auditorium that removed the brown M&Ms from the green room is to Van Halen.

It is common to get pushback on any YIMBY proposal from people who will say that it isn’t just a supply & demand, Econ 101, problem. They will say, this is complicated. Just saying “build more” is an oversimplification. You need a bunch of other changes (which, by the way, will each tend to block supply, add market frictions, and create unintended consequences that will keep supply from solving the problem).

In a way, I have been committing this error with missing middle housing. I don’t generally oppose missing middle proposals, but I haven’t been particularly excited by them. My reaction has been, “Sure, missing middle isn’t a bad idea, but it isn’t that simple. It can’t fix the problem.” Figure 1 highlights the basic reasoning.

Figure 1

Look, even in its heyday, gentle density (2-4 unit buildings), accounted for maybe 5% of total construction. Today, adding 5% just ain’t gonna cut it. We need much more than that. Gentle density, missing middle, whatever you want to call it, might be a nice part of the puzzle, but it isn’t going to be the primary driver of new supply.

I am pulling back on that position.

An Idle Thought

What triggered it was just an idle thought on a Sunday morning.

I would argue that we need at least 10 to 20 million additional units to get back to some functional normal housing market, which is affordable across the country. That is much higher than even the highest estimates from major housing institutions, which usually range from 1 to 5 million units.

So, I’m sure that I sound outrageous when I say 10 to 20 million. (As I have pointed out before, I think, we are 5 million vacant units short of normal. Unless all of the other forecasters have discovered some new secret way to run a housing market with fewer vacancies than we had in the past, and they haven’t told me about it, the standard estimates are laughably understated.)

My outrageous call for supply would take a building boom that lasts years, if not decades. This will not be solved overnight. But, as I pondered my outrageousness, I also realized that even the supply I call for amounts to something like a 10% increase in the American housing stock. It would take years because we typically only increase our stock of homes by about 1% annually. But, still, 10%? What a nothingburger!

Imagine a typical city block with maybe 10 single-family homes. Imagine that over the next couple of years, one of those homes adds a back yard casita or an apartment over the garage. Would that level of change even be noticeable? I’m sure that the proposal to do that in every neighborhood would trigger countless contentious neighborhood meetings with opponents lined up at a microphone to express their concerns about neighborhood character. But, in total, an outrageous amount of that housing would amount to practically nothing, aesthetically.

Put me on Team Missing Middle!

And, this brought me back to Figure 1. If all of that “outrageous” building would normalize our stock of housing with such a moderate amount of “missing middle” housing, then why do I simultaneously think that missing middle won’t be a large quantitative part of the solution?

I shouldn’t!

This is a 40 year old problem. We have been piling up dry powder. Figure 2 shows how many additional homes we would have today if we simply continued building 2-4 unit housing at the same rate that we built it in the 1970s.

Figure 2

So, sure, if we match the 1970s rate of missing middle building by building about 150,000 units annually (compared to the current rate of about 13,000), that won’t amount to much.

But, this country is full of millions of neighborhoods whose development has been stunted for 40 years. We could build a lot more than 150,000 units annually. And the 4.5 million unit deficit is an underestimate, because the category is relatively arbitrary. What if we rediscovered neighborhood-building? What if there was a little 12 unit building on a corner here and there with a cafe, or a coffee shop, or an ice cream shop, on the ground floor? That’s actually normal, though it doesn’t seem normal to Americans in 2024.

Obviously, development will not be scattered evenly across the urban landscape. Obviously there will be a place for everything from new single-family neighborhoods in the exurbs to skyrise condos in city centers. But, if we’re thinking about the sterile urban housing landscape that 20th century zoning has left us as a blank slate to build on, maybe I’ve had it backwards. Maybe this will be the “missing middle” century, and more skyrise condos and entry level single-family neighborhoods can also be parts of the broader movement.

And, it all seems so obvious as I think about it, I am embarrassed to write it, because some readers are probably justifiably thinking, “It’s about time you figured this out, Tweedle Dee.”

What if that blue line in Figure 1 popped up to 500,000 units, annually. Obviously, there are many constraints that would have to lifted for that to happen. But, the capacity of our existing neighborhoods to hold them is nowhere close to a constraint.

It is a prime example of a central issue in zoning reform: If you upzone one neighborhood, the broad housing market barely notices while the the neighborhood is greatly changed. If you upzone everything, the broad housing market gets fixed and each neighborhood barely notices.

One way we frequently get sidetracked is that the latter case is the YIMBY case, but American governance continually puts us in the former case. Maybe the question of how transformative missing middle housing can be is just a form of that problem.