The Real Reason For America’s Housing Crisis
What happens when, over the course of sixty years, the same number of people require 32 percent more housing units?
Source: gettyimages.com
Back in September, I wrote a piece that highlighted a notable but underestimated accomplishment in Chicago’s housing market. The 2023 U.S. Census’ American Community Survey data showed that Chicago reached its highest number of occupied dwelling units, 1.18 million, surpassing its previous high mark of 1.16 million in 1960.
On the surface this seems like an unremarkable feat. Yet this single data point might tell us more about the impact of demographic change in cities, and how such change alters their social and economic profiles, than anything. More from that article:
“The number of occupied dwelling units in Chicago hit bottom in 1990, with 1.03 million. Since then Chicago’s been adding about 4,500 occupied units annually, or about 0.4% each year.
It’s modest but meaningful change. It’s actual growth, really, especially considering that Chicago’s population dropped by 900,000 between 1960 and 1990, and an additional 150,000 since 1990. As Hertz says in the post, “(i)n other words, >100% of the population decline of the city is now explained by (decreasing) household size.””
You can see how much average household size decreased in Chicago in this chart produced by Jeremy Glover (@jgrantglover.bsky.social) and published at Blue Sky. With his permission (thanks!) I’m showing his chart here:
Between 2010 and 2020, 63 of Chicago’s 77 Community Areas (groups of neighborhoods that Chicago’s kept statistical tabs on for more than a century) saw a decrease in average household size. Of the 14 Community Areas that did witness an increase in average household size, eight were among the highest valued and/or lowest average household size Community Areas in the city). The other six were likely the benefactors of Latino migration from inner Chicago neighborhoods to areas closer to the city’s boundaries. Elsewhere, average household size keeps falling.
Chicago’s population has been virtually flat since 1990, but the number of occupied dwelling units is increasing. That got me thinking: how much of our nation’s housing crisis – our outright housing shortage – can be attributed to decreasing household size?
My guess is far more than most urban observers recognize.
I touched on this point in a series I started earlier this year, and republished just last month. In it I spoke of unconventional causes of the housing crisis:
“The nature of housing demand has changed significantly in the last half century. The national shrinking of the typical household size means demand has dropped dramatically since the peak Baby Boom years of the late 1950s and early 1960s. Average household size in the U.S. in 1960 was 3.38 persons. By 2023 that number fell to 2.55 – a 25% drop. With smaller household sizes, the demand for housing has increased relative to our overall population gains. This has led to a two-fold problem: 1) the housing industry has continued building the kinds of large single-family homes that were most in demand for decades, and 2) the housing industry has been late to respond to rising demand for multifamily rental properties (which lagged in construction until the last 3-5 years or so). They’ve begun to fill the void but haven’t yet succeeded.”
Unfortunately I didn’t follow up on that point in the series. I’m addressing that now.
Evaluating Housing Demand
First, let’s look at how housing demand is typically determined. What metrics are generally used? In an article published last month, David Wessel of the Hutchins Center on Fiscal & Monetary Policy at the Brookings Institution examined common methods for estimated housing need. The Brookings article notes that the most common method, and the one most often noted in the media, looks at changes in vacancy rates. For example, if the vacancy rate (for owner-occupied and renter-occupied homes) at any given time is lower than an established “normal” vacancy rate, the number of units needed to achieve the normal vacancy rate is considered to be the housing shortage figure. If the current vacancy rate is above the established normal rate, there’s a housing surplus.
The article then goes on to explore the “housing production” method. This method involves calculating an estimate of annual housing production over a given period, and applies that figure to the analysis period. For example, if the U.S. produces an average of a million new dwelling units a year over a 30-year period, then enters a phase when maybe 900,000 units are produced annually, there would be a loss of 100,000 units entering the housing market annually. The National Association of Realtors estimate in the chart below used this method and came up with an estimated 5.5 million dwelling unit shortfall nationally.
Here’s how four organizations that track housing data arrived at very different figures (this is reproduced from the original article):
The Brookings piece goes on to explain how each organization’s method differs and how they arrived at their conclusions; I won’t go into that here. What is important, however, is that over time the actual numbers may differ, but they do trend together.
What about decreasing household size?
Here’s the issue. I’m no economist, but it’s clear to me that the vacancy rate method is a demand-oriented methodology, while the production method is a supply-based one. In my view, the vacancy rate methodology is closest to determining whether there’s a housing shortage or glut nationally, and could be used effectively state, metro area and city levels. But I would say it still lacks the ability to factor variables that contribute to vacancy like location, obsolescence, segregation and more.
Meanwhile, the production method offers a good view of the housing industry’s capacity to produce housing, but very little about how much housing is actually needed. It reminds me of the Goodyear signs in Detroit during my childhood. They would tick off the number of cars produced by the Big Three automakers over the course of a calendar year, resetting each January. The signs were taken down in 2002 after years of falling figures. It became clear that while Americans were still buying cars, they weren’t necessarily buying American cars. With each year the signs became less useful as a measure of Detroit’s economic strength.
Neither methodology incorporates the national trend of decreasing household size into their analysis.
My point is this – decreasing household size has accelerated the demand for housing, even as population growth has slowed, and that acceleration has hardly been accounted for. Back in 1960, it’s assumed that 100 people could be accommodated by 30 homes (100 people/3.38 people per household = 29.6 homes). In 2023, a new generation of 100 people would want 39 homes (100 people/2.55 people per household = 39.2 homes). That means housing demand increased 32 percent, even with the number of people remaining equal.
Taking this point a little further, I’d say the nation is still overproducing homes for the traditional family arrangement that characterized American post-WWII society, and underproducing homes for the smaller families, singles, couples, and others who want homes scaled to their needs. I mean, that’s at the heart of NIMBYism and the YIMBY response, right?
Back in June, Zillow came closer to factoring decreasing household size into the equation. In June, Zillow issued a press release showing that their methodology estimates the number of “missing households” – persons living with other people that would prefer living on their own – and came up with a justifiable figure:
“At any given time in 2022, there were about 3.55 million vacant homes available for sale or for rent. At the same time, there were approximately 8.09 million individuals or families living with non-relatives, potentially preferring to live on their own – a difference of more than 4.5 million so-called “missing homes.”
“Despite an uptick in homebuilding the shortage of available homes continues to grow, mainly because of the increase in the nation’s family count and very sluggish construction activity in the decade preceding the pandemic. When new households are formed faster than the increase in the housing stock, the share of vacant homes falls, putting upward pressure on prices and rents.”
Maybe the auto industry analogy I made doesn’t resonate with everyone, but I think it makes sense. Just as there was a time when Detroit was overproducing Chevys when the nation was transitioning to buying Toyotas, the nation is still overproducing 4-bedroom, 3,000 square foot homes as the nation transitions demands smaller, more adaptable and more flexible living spaces. The difference is, Detroit’s Big Three had to adapt, go through the fire and come out on the other side. In the next few years that will be the challenge or the nation’s suburbs.
And please note during peak Baby Boom.... you had larger families often in smaller houses.
I am relieved to have this discussed and not ignored. We need to live together again, not build millions of tiny boxes for lonely people.