I want to dive a bit deeper on a point I made in yesterday’s post about the 2025 update to “The State of the Nation’s Housing” from Harvard’s Joint Center for Housing Studies.
The report notes, “While new construction has grown the rental supply overall, it has failed to increase the stock of low-rent units, which has dwindled over time. Between 2013 and 2023, the number of units renting for less than $1,000 per month after adjusting for inflation fell by more than 30 percent, from 24.8 million to 17.2 million. Instead, new construction focused primarily on higher-rent units, which helped to nearly triple the stock that rents for $2,000 or more, from 3.6 million to 9.1 million during the same period.”
…Housing doesn’t get more expensive because developers build more expensing housing… They claim that the number of apartments renting for $2,000 or more increased by 5.5 million units over a decade. There were less than 3.5 million apartment units, total, completed over that decade. The problem isn’t that we built too many “high rent” units.
The number of apartments renting for less than $1,000 per month (adjusted for inflation) declined by 7.6 million units. What happened to those units? Did they disappear? (Not more than a few.) Did they get nicer? (Some, but again, not more than a few.)
No. In almost all cases, apartments renting for, say $900 in 2013 now rent for, say, $1,200 (adjusted for inflation). And, what that means, in practical terms, in cities across the country, is that the cost of the political right to live in an apartment on the few plots of lands where it is legal, increased by $300 while the apartment stayed the same, or depreciated. The land increased in value by $300/month and the apartment stayed the same.
So, how are we supposed to build new $900 apartments? We would have to spend $300/month for land in order to build $600 apartments and charge $900 for them.
You might suggest that we should subsidize $900 apartments by using public money to pay for the $300/month land.
But, new $2,000 apartments can lower the scarcity value of land in a city as well or better than $900 apartments can. Why subsidize $900 apartments if we can get the same outcome by letting private builders build more $2,000 apartments?
And, certainly, why would you subsidize $600 apartments renting for $900 with one hand while you’re blocking $2,000 apartments with the other hand?
You might answer that there are market segments. There are $900 apartments and $2,000 apartments, and they house different families.
My question to you is: When the $900 apartments rose to $1,200, did they switch out families? Did those apartments switch to a different segment?
Some of them did. That is part of what has to happen when housing is politically obstructed. Families with the ability to pay $1,200 start moving into apartments that used to rent for $900.
But, many of those apartments have the same tenants as before. And, as I noted, the rents didn’t go up because they got nicer.
Market Segmentation
Actually, market segmentation is mostly a product of a functional, stable market. Back when we had those, you could believe that market segments were a state of nature as long as markets were functional.
Cities have blocked new apartments so thoroughly that apartment units now regularly transition to prices that should be associated with other market segments.
Under these conditions, market segmentation is weak, and every conflict in housing politics is a product of that. Every zoning rule. Every public meeting. They are all freak out sessions because market segments are mixing – apartments going in where single-family homes are nearby, workforce housing near luxury housing, single-family homes being torn down to build multi-plexes, homes being transitioned to duplexes, gentrification, etc.
All the efforts to block those forms of segment mixing have left homes no alternative but to just move to new segments in their current forms. Back in the before times, in functioning markets, market segments were pretty stable. $900 apartments didn’t become $1,200 apartments when adequate supply was legal.
What actually happened was that market segments mixed very slowly. Homes depreciate. In some neighborhoods, families reinvested, maintained, and upgraded. New buyers were buyers who wanted to move into neighborhoods that had been kept expensive.
Other neighborhoods depreciated. New buyers were buyers who wanted to move into neighborhoods they could afford.
The natural evolution of the geography of growing cities determined which neighborhoods were which.
Those differentiations operated at a pace slower than the natural propensity of houses to depreciate. That didn’t generally bother people. Sort of like how New York City has been expensive long enough (but not really for that long) to just seem like it should be expensive while when Phoenix home prices nearly doubled from 2004 to 2005, it really bothered everyone.
There’s a similar issue in market segmentation. Nobody is bothered by year after year of depreciation and house by house decisions about how thoroughly to maintain or upgrade the properties.
We don’t mind continuous functions. We don’t like discontinuous functions.
The most infamous discontinuous function on segment mixing led to white flight. A black family replacing a white family was a noticeable step-change, and many mid-20th Century whites didn’t like it.
A unit on a block full of homeowners being switched to a rental is a step-change.
A single-family unit being turned into a six-plex is a step change.
So, we had this pretty decent housing economy going with mostly continuous changes and a few neighborhoods on the margins with step-changes. For a while, of course, it was unfortunately legal to block the racial step-change. Zoning blocked the six-plex step change. Today cities are starting to block .
So, today, segments are mixing while the homes stay the same. $900 units are becoming $1,200 units.
Back when step-changes were legal, there was a whole lot of continuous change in American neighborhoods, and the occasional step-change.
Now that we have blocked most of the step-changes, all we are left with is the continuous change. But, now, the continuous change turns $900 apartments into $1,200 apartments instead of turning them into $600 apartments.
I have come to the conclusion that the bulk of excess real estate value in the United States is due to that horrible situation. Families don’t trade down easily. We don’t like mixing segments, especially when that means moving ourselves to a lower segment. The most effective way to turn a $30 trillion residential real estate market into a $40 trillion residential real estate market is to turn $900 homes into $1,200 homes while the families are stuck in them. That’s what we’ve done.
Frankly, at this point of my research, I don’t think low mortgage rates, credit booms, agglomeration economies, or the 20 other things that the Financial Crisis Inquiry Commission fingered in the pre-2008 housing boom could come close to the same scale of effect.
By eliminating all market segment mixing, we have put every single home on a market segment treadmill. Except, it isn’t improvements that are making them more expensive. They just sit on ever-inflating land.
The answer to that problem isn’t to block new homes for being too nice while building a limited number of new $900 homes with $300 rents for the land.
A Thought Experiment
Let’s say that instead of building 3.5 million apartments at, say, $2,000 rents, the US had built 3.5 million apartments at $900 rents. How would things be different? Would there have been more rent inflation or less? Would there have been more displacement or less?
First, let’s ignore the fact that those $900 rent apartments, which would either need to be $600 apartments on $300 land, or would need to be subsidized, would definitely be encroaching on some precious market segment, where $600 apartments or subsidized apartments definitely would not meet the approval of the existing neighbors.
Who would have moved into those apartments?
In the market we have, new families move into the $2,000 apartments, but there aren’t enough of them, so some families go pay $1,900 for formerly $1,800 apartments, and that demand keeps working its way down the housing stock, until you’re in neighborhoods where the existing tenants are holding on for dear life, spending $1,200 to try to stay in their $900 apartments.
If you add 3.5 million units at the bottom of the market instead of at the top, then, all you’re doing is forcing an additional 3.5 million families to trade themselves down to a lower market segment. And, as I argued above, we hate that. We hate it so much, we commonly are willing to pay $1,200 for $900 apartments to avoid it.
And, constrained supply conditions mean that rent inflation over time is unrelated to the condition of the units. If the total number of units built was still 3.5 million, then rents will rise just the same on all those units. (Actually, I would argue that demand for housing is inelastic, and if we built 3.5 million $900 units instead of 3.5 million $2,000 units, after American families sorted themselves into them, we would be paying more in total to live in them. For the past 50 years, the less we invest in new housing over time, the more we pay for it in total.) They would be back to living in $1,200 units, but now, in order to distribute themselves into those units, 3.5 million families would have had to trade down to the new units so that other families could trade down into their units.
You can throw money at it, or subsidies some families, or freeze rents. That doesn’t change this process. Selective subsidies and price controls just squeeze the balloon so it bulges out somewhere else.
Under current conditions, building high-rent homes instead of low-rent homes reduces displacement, because the current balance of household moves is a combination of families moving down into the low-tier and families in low-tier units spending more to stay in place.
Building enough high-rent units reduces the number of downward compromises and eventually allows some aspirational housing changes so that families get to trade up or they get to stay in the homes they already live in while rents decline.
City-Building
You could think about the transitions from unzoned cities to zoned and mortgage-suppressed cities in these terms.
In unzoned cities, there was some margin where step-change segment mixing happened. Somewhere near the city center, occasionally, there was a neighborhood that saw its first townhome or side unit or multi-plex. Then, that neighborhood would proceed along its new path of continuous change, slowly densifying until it was full of apartments or townhomes. There are only a few cities where the remnants of that process remain.
When that step-change mixing was outlawed, most cities could overcome it by building homes in the suburbs. So, instead of having well-located, dense neighborhoods, cities mostly just had aging single-family homes. This failed to serve a lot of needs that traditional cities used to serve, but most cities were able to remain affordable.
Depreciating single-family homes replaced dense, connected homes where dense homes had been outlawed. Twentieth-century American cities had continuous change rather than having mostly continuous change with occasional step-change targeted to match the changing proclivities of newcomers into the centrally located neighborhoods. Then, the mortgage crackdown changed the direction of that continuous change. Now, depreciating homes sit on inflating lots.
That’s where we find ourselves. Less valuable structures are not a remedy for inflated land.