Every month, the Census releases new numbers on new housing construction. As an indicator of current economic conditions, June’s numbers didn’t give any dramatic news one way or another. But they did highlight a trend that I think should get more attention: the decline of single-family housing in the US.
To market watchers, housing is an important sign of business cycle turning points. A well-known article argues that Housing Is the Business Cycle. From this point of view, June’s numbers were not very informative. They told the same story the last several months’ did: After steadily rising from the end of the recession, housing construction has stabilized — housing starts and permits issued have been basically unchanged since early 2017. Last month’s housing starts were almost exactly the same as last summer’s. The fact that housing construction is no longer rising might perhaps be seen as a sign of economic weakness; but it’s hard to take it as a sign of a crisis or imminent downturn.
But pulling back from the month by month variation, the most recent numbers reflect two related trends that may be more important than the ups and downs of the business cycle.
The first trend is the secular decline in housing construction. Housing starts, while higher than a few years ago, are still very low by historical standards — not just compared with the boom period of the 2000s, but with most earlier periods as well. On a per capita basis, new housing construction is at a level seen only at the bottom of the worst recessions before 2007. Compared with an annual average of 6.5 new units per thousand people in the 1980s and 1990s, the current rate is less than 4 per thousand, and shows no sign of returning to the old rate.
It’s hard to say how much this decline in new housing construction is a specifically post-bubble-and-crisis phenomenon, and how much it reflects longer-term trends. People sometimes suggest that low rates of housing construction are the flipside of the housing boom of the 2000s. There was a strong case for this in the years immediately after the recession, when the fraction of vacant houses was well above historical levels. But since then, the inventory of vacant houses has come down toward more normal levels.
Meanwhile, if we look at new housing construction per capita over a longer period, there is a fairly steady long-term decline – it’s not clear that the most recent period is exceptional. If you draw an exponential trend from 1959 through 1999 (the start of the housing bubble), as shown in the figure below, the current level of housing starts falls right on that trend. And relative to the shortfall in new construction during 2008-2015 is not too much greater than the excess of new construction during 1999-2007. To put it another way, the percentage decline in housing starts per capita over the past 20 years, is not much bigger than the average decline over any 20 year period since the 1950s.
Of course, this is just one way of looking at the numbers. There are many ways to draw a trend! And one might argue that, historically, the top of a boom should see new housing starts well above trend, suggesting that the recent decline is something new after all. You might also reasonably wonder whether the long term trend has any substantive meaning at all. The political economy of housing the 1950s and 1960s was different from today on all sorts of levels. It wouldn’t be hard to look at the same data in terms of a structural break, rather than — or in addition to — a downward trend.
For macroeconomic purposes, though, it doesn’t necessarily matter. Whether it reflects the ongoing effects of the subprime crisis or whether it reflects longer-term factors — slowing population growth, an aging population, the end of suburbanization – or whether it’s some mix of both, the decline in new housing construction remains an important economic fact.
Among other things, it is important for macroeconomic policy. Mortgage lending is central to the financial system: Housing accounts for over 70 percent of household debt, and housing finance plays a central role in financial instability. Conversely, residential construction is the economic sector most sensitive to financial conditions, and to monetary policy in particular. So the shrinking weight of housing in the economy may be a factor in the Federal Reserve’s inability to restore growth and full employment after the crisis. Looking forward, if conventional monetary policy works primarily through residential construction, and residential construction is a permanently smaller part of the economy, that is another argument for broadening the Fed’s toolkit.
Housing construction may be down for the count, at least compared with historical levels. But — and this is the second trend – it is not down across the board. The recent decline is limited to single family housing. Multifamily construction has been quite strong, at least by the standards of the post-1990 period. Compared with the two decades before 2007, single-unit housing starts in the past year are down by a third. Multifamily starts are up by a third. Per capita multifamily housing starts are actually higher than they were at the height of the housing boom. 
These divergent trends imply a major shift in the composition of new housing. Through much of the 1990s, less than 10 percent of new housing was in multifamily projects. Today, the share is more like 30 percent. This is a dramatic change in the mix of housing being added, a shift change visible across much of the country in the form of suddenly-ubiquitous six-story woodframe apartment buildings. The most recent housing data released suggests that, if anything, this trend is still gathering steam: A full third of new housing in June was in multifamily buildings, an even higher proportion than we’ve seen in recent years. In the areas that the Census designates as metropolitan cores, the shift is even more dramatic, with the majority of new housing units now found in multifamily buildings.
The shift in new construction away from single-family houses is consistent with the decline in homeownership. At 64 percent of households, the share of homeowners is 5 points lower than it was in the mid-2000s. In fact it’s back almost exactly where it was 30 years ago, before the big expansion in homeownership of the 1990s and 2000s.
To be sure, multifamily housing and rental housing are not the same thing. But there is a very substantial overlap. Over 80 percent of detached single-family homes are owned by their occupants. Less than 20 percent of units in larger buildings are, and the share drops as the number of units in the building rises. While homeownership rates have fallen across the board over the past decade, these relative patterns have not changed. (See the figure below.) So it’s fair to say that the decline of homeownership and the shift toward multifamily developments are, if not the same trend, at least closely linked.
The aggregate figures understate the decline in homeownership, because over this period the population has also been aging, and older families are much more likely to own their homes. (For a good discussion of these trends, see here.) For younger families, homeownership rates are lower than they have been in many decades. Compared with 40 years ago, homeownership rates are substantially lower for every age group except those 65 or older. Even compared with a decade ago, there has been a substantial fall in homeownership rates in younger age groups. As a result, the typical homeowner today is much older than in the past. Only a quarter of US homeowners today are younger than 45, compared with nearly half in the 1980s.
The same pattern is visible over the post-housing crash period, as shown in the figure below. Among those aged 30-44 – the ages when most Americans are starting families – the rate of homeownership is nearly 10 points lower than it was just a decade ago. The shift in housing construction toward multifamily buildings reflects the fact that Americans in their prime working years are much more likely to be renters than they used to be. 
This shift is important for politics as well as the economy. Tenant organizations were once an important vehicle for mass politics in American cities. In the progressive imagination of a century ago, workers were squeezed from one side by landlords and high rents just as they were squeezed from the other by bosses and low wages.
After World War II, the focus of housing politics shifted away from tenants’ rights, and toward broadening access to home ownership. This shift reflected a genuine expansion of homeownership to middle class and working class families, thanks to a range of public supports — supports, it should be noted, from from which African-Americans were largely excluded. But it also reflected a larger vision of democratic politics in terms of a world of small property owners. Homeowners were expected — not without reason — to be more conservative, more ready to imagine themselves on the side of property owners in general. As William Levitt, developer of the iconic Long Island suburb, is supposed to have said: “No man who owns his own house and lot can be a communist.”
The idea of a property-owning democracy has deep roots in the American political imagination, and can be part of a progressive vision as well as a conservative one. Baby bonds – an endowment or grant given to everyone at the start of their life — are supposed to be a way to broaden property ownership in a way that opens up rather than shuts down possibilities for radical change. Here for example is Darrick Hamilton in his 2018 TED Talk. “Wealth,” he says,
is the paramount indicator of economic security and well-being. It provides financial agency, economic security… We use words like choice, freedom to describe the benefits of the market, but it is literally wealth that gives us choice, freedom and optionality. Wealthier families are better positioned to finance an elite, independent school and college education, access capital to start a business, finance expensive medical procedures, reside in neighborhoods with higher amenities… Basically, when it comes to economic security, wealth is both the beginning and the end.
Descriptively, there’s certainly some truth to this. And with homes by far the most important form of middle-class wealth, policies to promote homeownership have been supported on exactly these grounds. Homeowners enjoy more security, stability, a cushion against financial setbacks, and the ability to pass their social position on to their children. The policy problem, from this point of view, is simply to ensure that everyone gets to enjoy these benefits.
One way to keep people secure in their homes is to allow more people to own them. This has been the focus of US housing policy for most of the past century. But another way is to give tenants more of the protections that only homeowners currently enjoy. Outside a few major cities, renting has been assumed to be a transitory stage in the lifecycle, so there was little reason to worry about security of tenure for renters. A few years ago I was a guest on a radio show on rent control, and I suggested that apart from affordability, an important goal of rent regulation was to protect people’s right to remain in their homes. The host was genuinely startled: “I’ve never heard someone say that a person has the right to remain in their home whether they own it or not.”
There are still plenty of people who see the decline in homeownership as a problem to be solved. But the shift in the housing stock toward multifamily units suggests that the trend toward increased renting is unlikely to be reversed any time soon. (And even many single-family homes are now owned by investors.) The experience of the past 15 years suggests that, in any case, home ownership offers less security than we used to think.
If more and more Americans remain renters through their adult lives, the relationship with the landlords may again approach the relationship with the employer in political salience. Strengthening protections for tenants may again be the basis of political mobilization. And people may become more open to the idea that living in a place, whether or not you own it, gives you a moral claim on it — as beautifully dramatized, for example, in the 2019 movie The Last Black Man in San Francisco.
We may already be seeing this shift in the political sphere. In recent years, there has been a resurgence of support for rent regulation. A ballot measure for statewide rent control failed in California, but various bills to extend or strengthen local rent regulation have gotten significant support. Oregon recently passed the nation’s first statewide rent control measure. And in New York, Governor Cuomo signed into law a sweeping bill strengthening rent regulation where it already exists — mainly New York City – and opening the way for municipalities around the state to pass their own rent regulations.
The revival of rent regulation reflects, in the first instance, political conditions – in New York, years of dogged organizing work by grassroots coalitions, as well as the primary defeats of most of the so-called Independent Democratic Conference, nominal Democrats who caucused with Republicans and gave them control of the State Senate. But it is not diminishing the hard work by rent-regulation supporters to suggest that the housing-market shift toward rentals made the terrain more favorable for them. When nearly half the population are renters, as in New York State, there is likely to be more support for rent regulation. The same dynamic no doubt played a role in the opposition to Amazon’s new headquarters in Queens: For most residents, higher property values meant higher rents, not windfall gains.
To be sure, the United States is not (yet) New York. The majority of American families still live in homes they own. But as the new housing numbers remind us, it’s a smaller majority than it used to be, and likely to get even smaller in the future. Which suggests that, along with measures to democratize property-ownership, there is a future for measures like rent control, to ensure that non-property owners also have a secure claim on their part of our common wealth.
(Figures 1, 3 and 4 are my analysis of series from FRED: HOUST, HOUST1F, COMPUTSA, and POPTHM. Figure 2 is from the Census Housing and Vacancy Survey. Figures 5 and 6 are my analysis of ACS data.)
