An analysis of population trends in urban vs. suburban style U.S. communities.
Where is the U.S. growing fastest, in cities or suburbs? And within those broad categories, what kinds of neighborhoods are growing fastest? Knowing where population growth is taking place is essential for projecting demand for infrastructure and public services. It can also be a window into what kinds of communities people prefer, though a cloudy one. Where people actually settle depends on much more than what they say they want. Housing supply, energy prices, tax policy, and other factors all affect the availability and cost of housing and, as a result, where people end up living.
In recent years, pinpointing where the U.S. population is growing has been particularly contentious, tangled up in the housing bubble and its aftermath. The housing bubble fueled construction and population growth in the suburbs and beyond, while, at the bubble’s height, the most urban counties actually lost population in absolute terms. After the bubble burst, growth returned to the highest-density counties, and, in 2011, the highest-density quartile of counties was the fastest growing. In the post-bubble years (2006-2013), urban population growth accelerated while suburban population growth slowed compared with the bubble and its run-up (2000-2006). That lends support to the view that America was seeing a permanent shift toward more urban living.
But things look different if we consider the entire cycle to date (2000-2013). Over that span, the densest quartile of counties was the slowest growing, just as in the two previous decades (1980-2000). And the fastest-growing large metros since 2000 are mostly the same ones that led in the 1980s and 1990s. All this casts doubt about whether there’s been a permanent, longer-term shift in population growth patterns. (Data supporting the above points are here.)
Measures of city versus suburban growth suffer from both conceptual and data limitations, and these are related. Conceptually, city boundaries don’t always line up with the qualities we associate with urban places. Many places outside big-city boundaries, like Hoboken, Cambridge’s Central Square, and Santa Monica, are denser, more walkable, and better served by transit than some neighborhoods within city boundaries, like parts of Staten Island and the western San Fernando Valley in the City of Los Angeles. Analyses based on city or county boundaries can’t tell us whether more growth is occurring in neighborhoods that are urban or suburban in nature. The Census publishes annual population estimates for counties, cities, and towns. But that doesn’t reveal whether the City of Los Angeles is growing more downtown or in West Hills, 30 miles away.
An alternative data source overcomes some of these limitations. The U.S. Postal Service reports monthly the number of residential addresses receiving mail in each ZIP code. This gives us a more granular, up-to-date view on changes in the number of households. Meanwhile, the Census Bureau publishes data on ZIP code (technically, ZIP Code Tabulation Area, or ZCTA) characteristics, like housing type and household density. This helps identify neighborhoods that are more or less "urban" or "suburban" in how they are designed and experienced, regardless of official city boundaries. It also allows us to compare suburbs that are more dense or diverse with those that are less so. In this post and in related work published on the Trulia Trends blog, I combined household growth estimates derived from USPS data with urban-suburban definitions based on neighborhood characteristics gleaned from Census ZCTA data.
Housing density is an intuitive way to classify neighborhoods as urban or suburban. Let’s say we classify neighborhoods in the 100 largest U.S. metros where most housing units are detached single-family homes as suburban. That would mean classifying those where most housing units are in multi-unit buildings or attached homes as urban. (This definition has the theoretical advantage over a traditional density measure like the number of housing units or households per square mile in that it’s not skewed by open space, commercial areas, or other non-residential uses within a ZCTA. In practice, however, using households per square mile yields similar conclusions.) For example, this definition properly categorizes Hoboken as urban and West Hills, California, as suburban — the opposite of how city boundaries classify those areas. Moreover, this definition of urban and suburban based on housing density lines up better with residents’ perceptions of their neighborhood than a classification based on big-city boundaries, according to a recent Trulia survey.
As we reported elsewhere using these data and definitions, suburban neighborhoods modestly outpaced urban neighborhoods in population growth (technically, household growth) in 2014, 0.96 percent to 0.85 percent. Since mid-2011, suburban neighborhoods have grown 3.1 percent and urban neighborhoods 2.5 percent. But the suburb category embraces a wide range of neighborhoods. Which kinds of suburban neighborhoods have grown fastest? To see, let’s differentiate suburban neighborhoods based on measures of the degree of urbanness.
We used six different, though correlated, measures to divide suburban neighborhoods into a relatively more urban half and a relatively less urban half. These include the share of detached single-family homes; household density; income diversity; racial and ethnic diversity; prevalence of commuting by transit, walking, or biking; and restaurant density. To separate the suburban neighborhoods into a more urban half and a less urban half, we calculated the median value of each urbanness measure among suburban neighborhoods as defined above in the 100 largest metros, weighted by the number of households.
|Measure of urbanness||(1) Median value of the urbannness measure, across all suburban neighborhoods||(2) Household growth since mid-2011 in the more urban half of suburbs||(3) Household growth since mid-2011 in the less urban half of suburbs|
|Share of housing units in multi-unit buildings and attached single-family homes||26%||2.4%||3.9%|
|Household density per square mile||695||1.8%||4.4%|
|Income diversity (Gini coefficient)||.41||2.4%||3.8%|
|Racial/ethnic diversity (share of residents outside of neighborhood’s largest racial/ethnic group)||25%||3.4%||2.8%|
|Percent of workers commuting by transit, bicycle, or on foot||3.4%||1.9%||4.3%|
|Restaurants per square mile||2.5||2.0%||4.2%|
Note: this table shows growth in the more urban half (column 2) versus less urban half (column 3) of suburbs for six different ways of dividing suburbs in half (third and fourth columns). For each measure, the more urban half of suburbs are those whose value of the measure is above the median (column 1) across all suburban neighborhoods. These measures of urbanness come from the decennial Census, American Community Survey, or ZIP Code Business Patterns — all produced by the U.S. Census Bureau.
For every measure but one, household growth in 2014 was faster in the less urban half of the suburbs. The one exception was racial and ethnic diversity. Suburbs with a higher share of residents outside the neighborhood’s largest racial or ethnic group had faster growth. This is a reflection of the fact that growth nationally is faster in metros with greater racial and ethnic diversity. But suburbs that are less diverse relative to other suburbs in their own metro grew faster than suburbs that are more diverse (technical note: based on a regression of growth on diversity with metro fixed effects).
In short, suburbier suburbs have recently been growing faster than more urban-style suburbs.
To be sure, dividing suburbs into halves might overlook finer-grained differences. Furthermore, the definition of suburbs as neighborhoods where a majority of housing units are detached single-family homes is itself arbitrary. An alternative is to treat urbanness and suburbanness as a spectrum. To get a view of this, we combined our six measures of urbanness into a single score for each neighborhood. The more urban the neighborhood, the higher the score. (Technical note: the score is the first factor from a principal components analysis of our six measures. The four density and commuting measures ended up contributing more to the score than the two diversity measures. Replicating the analysis that follows without the diversity measures does not change the results materially.)
Using this score, we divided all neighborhoods in the 100 largest metros into deciles with approximately equal numbers of total households, from the most urban to the most suburban. Urban neighborhoods, those where most of the housing is not single-family detached homes, correspond roughly to deciles 1 – 3; the relatively more urban suburbs are deciles 4 – 6; and the relatively less urban suburbs are deciles 7 – 10. This chart shows growth since mid-2011 in each decile.
The results of this analysis are striking. In recent years, the fastest-growing neighborhoods are the most suburban. These are the lowest density, most car dependent, overwhelmingly single-family neighborhoods. Our decile breakdown confirms that the fastest growing suburbs are the less urban suburbs, not the more urban ones.
However, that’s not the end of the story. Among urban neighborhoods, the most urban (decile 1) have grown fastest. In fact, this most urban tenth of neighborhoods has had faster growth than not only other urban neighborhoods (deciles 2 and 3), but also deciles 4 through 7, the more urban suburbs. This most-urban decile of neighborhoods includes essentially all of Manhattan, Brooklyn, and the Bronx; most of Queens and the cities of San Francisco, Boston, and Washington, DC. It also includes some neighborhoods like Hoboken that lie outside big-city boundaries. Deciles 2 and 3 – urban but not hyper-urban – are the most common neighborhoods in Los Angeles and Miami-Dade counties, though both counties also include many urban neighborhoods in the top decile.
These national patterns are consistent across the four Census regions. In all regions, the less urban suburbs are growing faster than the more urban suburbs. The fastest growing neighborhoods nationally are the lowest-density suburbs in the South and West. At the same time, in all four regions, the densest urban neighborhoods are growing faster than other urban neighborhoods and many suburban neighborhoods.
So are suburban areas growing faster than urban areas? The simple answer is yes. But the fuller answer is that some urban neighborhoods are growing fast and some suburban neighborhoods even faster. The best evidence of urban growth is in the densest city neighborhoods, not in a shift within suburbia toward more urban suburbs. Growth is currently favoring the densest urban neighborhoods and the most suburban suburbs, not the neighborhoods in between.