(Re-)Mapping America's Metropolitan Growth: Islands of Density in a Sea of No Growth
Editor’s Note: The expensive coastal cities’ current housing affordability crisis has been decades in the making, the result of slowing metropolitan growth in the face of sustained housing demand. This piece explores the historical housing development patterns in a series of different metros nationwide, and illustrates how each is representative of larger groups of similar markets. For a deeper dive into potential new housing that could be developed under a variety of densification scenarios should current restrictions on density be relaxed, we encourage you to read our additional research here.
There are currently almost 140 million homes in the United States, and analyzing where, when and how they were built tells a vivid story of America’s metropolitan growth. But more than a history lesson, having a clear view of our past development patterns is vital for understanding the causes of our current housing affordability issues, and for shaping effective solutions.
Cities and metro areas grow in two ways: By expanding outward into the surrounding, undeveloped countryside; and/or by growing denser within an already-developed footprint. Historically, American cities have done both. But since the 1970s, several of the country’s most economically vibrant coastal cities have gradually been doing less and less of either. The result has been decades of faster-than-average housing price growth in these areas as they’ve struggled to add sufficient new housing in the face of sustained housing demand, which underpins their current housing affordability crisis.
Additionally, residential development in essentially all American metros has shifted to a new pattern. Construction in metropolitan interiors now concentrates in clusters of very large apartment buildings, dispersed amid mature, less-densely developed suburban tracts that are virtually frozen in time and produce almost no new housing. There is enormous potential for new housing supply to be had by allowing for even modest amounts of new density – allowing a small share of lots currently host to a single-family home to instead house two, three or four units – in these otherwise stagnant areas.
Mapping Growth
The following maps allow us to visualize the growth of large U.S. metros over the past 80-odd years, detailing what type of housing has been built, where and when over the decades.
Here’s how to interpret these visualizations:
The colors reflect the mixture of housing types: Single-family homes in blue; small and medium-sized apartment buildings (2 to 49 units) in yellow; and large apartment buildings (50 units or more) in red.
The degree of transparency corresponds to housing density – more transparent means less dense, more opaque means more dense
Note: When viewing housing from a specific timeframe, e.g. from 1960-1980, the map will appear more transparent because the opacity corresponds only to the incremental housing density built during that period. Viewing multiple vintages overlays the transparencies, illustrating how certain areas have become gradually more dense over time.
With these visualizations in mind we can consider how three very different metropolitan regions have grown over time, and illustrate how each is representative of larger groups of similar regions.
Los Angeles
The Los Angeles region expanded into the surrounding countryside very quickly in the second half of the 20th century, but its pace of expansion substantially slowed after the 1980s. From 1940 to 1990 the region’s developed footprint expanded by 406 square miles per decade, on average. From 1990 to 2010 it expanded only 215 square miles per decade (and the pace in the 2010s has likely been even slower).
Densification – adding more homes to already-developed areas – was common as recently as the 1980s and ‘90s, often taking the form of small and medium-sized apartment buildings (2 to 49 units). But densification has largely tapered off, with most of the L.A. metro’s interior producing very little new housing. Residential construction in the interior is now increasingly concentrated in clusters of very large apartment buildings (50+ units). The visual signature of L.A.’s development pattern over the most recent 20 years – most aptly described as islands of density in a sea of no-growth – is a transparent metropolitan interior speckled with small clusters of red.
L.A.’s development pattern is typical of west coast metros including the San Francisco Bay Area, Portland, Seattle and San Diego. To a lesser extent, the Los Angeles region’s metropolitan growth also resembles that of metros in the northeast, including New York, Boston, and Washington.
Houston
The Houston region also expanded into the surrounding countryside very quickly in the second half of the 20th century. But unlike the Los Angeles region, Greater Houston’s expansion has not slowed down.
Because the region has accommodated new residents largely through outward expansion, there has been less buildup of demand for housing within the developed footprint, and therefore less densification than in L.A. and other areas. Indeed, the region’s outward expansion has provided a relative abundance of housing that has helped keep affordability in check and house prices themselves largely tethered to the cost of construction rather than just people’s willingness to pay.
Still, some popular areas in Houston’s metropolitan interior have seen substantial densification. Although it’s relatively easy for the metro as a whole to expand outward, areas within it are “landlocked” by surrounding developed areas – leaving growth through densification as their only option. When demand for living in these specific areas rises, they can only grow denser and/or more expensive.
As elsewhere, densification within Houston’s already-developed area has also gradually taken on the islands of density pattern.
The Houston region is typical of other growing and affordable metros in the south, including Atlanta, Austin, Dallas, Phoenix and Raleigh. To a lesser extent, it also resembles some metros in other parts of the country experiencing growth pressures but also surrounded by plenty of open land, such as Minneapolis, Denver and Sacramento.
Detroit
Detroit has had its share of troubles, but even so its developed footprint continues to expand. Still, the pace of that growth is modest compared to a place like Houston – or even Detroit’s own historical growth.
Given the lack of constraints on the region’s outward expansion, as well as the relatively weak demand for living there, the Detroit region’s metropolitan interior has experienced little densification. But again, as in other parts of the country, the densification that has occurred in recent years has shifted towards the islands of density pattern.
The Detroit region is typical of other Rust Belt metros including Buffalo, Cleveland, Milwaukee and St. Louis. To a lesser extent, it is also reflective of the Pittsburgh region and parts of the Philadelphia and Chicago metros.
Islands of Density in a Sea of No-Growth
Because of local rules and regulations that have accrued over the decades, it is rare today to see the nation’s low-density suburban tracts redeveloped more densely. Areas zoned for single-family homes are particularly well-insulated from such change, and rarely see the types of multifamily construction common as recently as a generation ago. Such areas account for most of metropolitan America’s land, and comprise the sea of no-growth.
That sea is punctuated by islands of density, almost the only places in which densification still occurs today. They consist of clusters of very large apartment buildings (50+ units), and they typically emerge in areas including:
Suburban downtowns, where density is grandfathered in
Near transit, where the case for density is most compelling
Formerly non-residential areas where their development draws less local opposition.
The Future
As America struggles with a housing affordability crisis, especially in the expensive coastal cities, reinvigorating outward expansion – in other words, sprawl – is generally regarded as unsustainable. This leaves densification as the main alternative.[1]
That raises an important question. Should densification be confined to islands of density, as it is today, or should American cities also come to accept modest densification of the vast existing sea of no-growth? The two options are complementary, not mutually exclusive, and pursuing them jointly could allow cities to produce a greater number of homes and a richer mix of housing types than they could otherwise.
Notes
[1] Another alternative is embracing smaller homes, as espoused by the movement around tiny homes and micro-units, but that appears to be a niche trend.
Data & Methodology
The data presented in the maps are drawn from the 2013-2017 5-year compilation of the American Community Survey and reflect the existing housing stock as of that period. For more information on the advantages and disadvantages of using the current stock of existing homes to indirectly observe past patterns of metropolitan growth, see the data and methodology sections of the first two studies in the Further Reading section.
The areas represented in the maps are U.S. Census tracts, and the extent covered by each map is determined by the current definition of the metro area, which is either a CBSA or a CSA where one exists. Tract hue is determined as a linear combination of the number of the number of housing units built during a period in each of the three structure size categories: Single-family, 2-49 unit and 50+ unit structures. Tract opacity is determined by the incremental housing density built during each period, which is subjected to the following transformation: opacity(x) = k -1log(1+x), where x denotes incremental housing density (in terms of homes per acre) for one or more periods and k = log(1+max{x}), where max{x} is the maximum observed cumulative density for the entire timespan across all Census tracts in the pool of metro areas mapped. The transformation maps into the range [0,1] and is monotonic. It ensures that virtually all tracts are partially transparent given any combination of displayed time periods. Only for periods during which incremental housing density was strictly zero is a tract be fully transparent, and only a single tract whose cumulative housing density across all periods is the national maximum is fully opaque when all time periods are displayed concurrently.