Population Growth in Dense U.S. Cities: Short-Term Correction or Long-Term Trend?

People are moving to city centers in record numbers, but that doesn't necessarily mean it's a permanent change in where Americans want to live.  

Image
Reuters

Has the housing bubble and bust fundamentally changed where Americans want to live? Today, apartment construction in dense cities is booming, and hyper-urban high-rise neighborhoods have both strong home-price growth and population growth. Furthermore, the most recent Census population estimates tell a striking story of a swing back to big, dense cities.

After losing population between 2000 and 2006, Brooklyn, San Francisco, Washington, D.C., and Suffolk County, Massachusetts (Boston) all grew between 2010 to 2013, and faster annually than between 1980 and 2000. At first glance, these trends suggest that the housing bubble and bust reversed a decades-long shift toward suburbs and the Sunbelt, harkening an urban resurgence.

But not so fast.

These recent trends look cyclical, not necessarily structural: that is, they reflect the current phase of the housing recovery and represent a correction to the housing bubble, rather than a deeper, underlying "structural" shift in population. The densest counties – like the four mentioned above – fared badly during the bubble but have benefited in the most recent phase of the cycle, as young adults hurt by the recession have begun to get jobs and move out of their parents' homes into urban apartments. Grouping counties by density shows how differently the housing cycle affected different types of places, with the bubble (2000-2006) favoring the lower-density counties and the recovery (2010-2013) favoring denser places:

 

To look for evidence of a more permanent, structural shift in population growth patterns, it helps to average out the cyclical trends by looking over this whole cycle, 2000-2013, compared with what came before. It turns out that the patterns of population growth in the 21st century are overwhelmingly a continuation of the last decades of the 20th century.

Among the 100 largest U.S. metro areas (including metro divisions, where defined, based on the 2009 government defintions), the fastest-growth metros are all in (or on the edge of) the Sunbelt:

#

Metro

Annual population growth, 2000-2013

1

Raleigh, NC

3.2%

2

Austin, TX

3.1%

3

Cape Coral-Fort Myers, FL

3.1%

4

Las Vegas, NV

2.9%

5

Charlotte, NC-SC

2.6%

6

Orlando, FL

2.4%

7

Phoenix, AZ

2.3%

8

Houston, TX

2.3%

9

Riverside-San Bernardino, CA

2.3%

10

San Antonio, TX

2.2%


... which looks very similar to the list of fastest-growing metros in the last two decades of the 20th century:

#

Metro

Annual population growth, 1980-2000

1

Las Vegas, NV

5.6%

2

Cape Coral-Fort Myers, FL

3.9%

3

Austin, TX

3.9%

4

Riverside-San Bernardino, CA

3.8%

5

Orlando, FL

3.6%

6

Phoenix, AZ

3.6%

7

Raleigh, NC

3.5%

8

West Palm Beach, FL

3.4%

9

Atlanta, GA

3.1%

10

Palm Bay-Melbourne-Titusville, FL

2.8%


Comparing the top-ten lists from 1980-2000 and 2000-2013, seven metros (in bold) appear on both. The changes were that West Palm Beach, Atlanta, and Palm Bay – Melbourne – Titusville were replaced by Charlotte, Houston, and San Antonio: a swap of three middle-to-lower-density Sunbelt metros for three others.

The persistence of population growth patterns goes beyond the top ten metros. The correlation between metro-level population growth in 2000-2013 and in 1980-2000 among the 100 largest metros is 0.87, and the scatterplot shows how tight the relationship is:

 

Switching from 100 large metros to all U.S. counties in order to capture all metropolitan and non-metropolitan areas, the story of continuity holds. The correlation between 1980-2000 and 2000-2013 growth among all U.S. counties (population-weighted this time) is 0.83. Furthermore, despite the turmoil of the housing bubble and bust, the persistence of population growth patterns has strengthened over time: the correlation between 1980-2000 and 2000-2013 population growth (at the county level) was higher than the same correlation between 1960-1980 and 1980-2000, and than between 1940-1960 and 1960-1980.

The persistence of metro- and county-level population growth patterns suggests that the long-term trends toward middle-density areas – suburbs rather than big cities or rural areas -- and toward the Sunbelt have continued. Indeed, they have.

Grouping counties into quartiles based on density (weighted density, explained here) shows that the highest-density quartile of counties was the slowest growing quartile, both since 2000 and between 1980 and 2000. Although, as shown in the first chart, above, the highest-density counties grew faster in the last year, 2012-2013, than the third quartile (density of 500-1000 households per square mile) and just a bit slower than the second quartile (density of 1000-2000 households per square mile), much of that reflects that the housing recovery has just hit the point when young people are starting to move out of their parents’ homes into urban apartments: in other words, a cyclical – not necessarily structural – trend.

The next chart shows that warmer counties, based on mean January high temperatures, grew faster than colder counties, both between 2000-2013 and between 1980-2000. The difference narrowed but didn't come close to reversing in favor of colder counties.

 

Overall, the latest population growth data show that the housing bubble, boom, and recovery have not fundamentally changed population growth patterns in the U.S. The most recent year-over-year changes do show that the housing recovery is, at the moment, favoring growth in many dense counties. But averaging over the whole cycle of 2000-2013 shows that the pattern of population growth favoring the middle-density counties (suburbs and smaller cities) and the Sunbelt remains largely unchanged from the last century.

The persistence of population growth patterns is a cautionary tale about claims that the housing bubble and bust have fundamentally changed where Americans want to or will live. While there is no doubt that apartment buildings are leading the construction recovery, homeownership is at its lowest level in almost 20 years, and many big cities are growing faster today than during the bubble, these facts all reflect a particular moment in the housing cycle and are, in large part, reactions to the dynamics of the bubble and bust.

If there have also been permanent, structural shifts in how and where Americans live, those shifts probably can’t be proven or disproven by recent trends. The extreme housing cycle of the 21st century has caused sharp short-term swings in home prices, construction, and even population – sharp enough to overshadow and obscure underlying structural changes that could be happening at the same time. The demographic, attitudinal, or public policy shifts that could fundamentally change how and where American live are likely to evolve more slowly than the ups and downs of the housing cycle. Trends based on a few years of data – especially from this recent housing cycle – should be presumed cyclical until proven structural, rather than the other way around.

About the Author