If you have trouble reading this email, please go to the online version.
March 21, 2018

The Blurring Line Between Cities and Suburbs

Locations within metropolitan areas have traditionally been divided into “cities” and “suburbs.” For many, the idea of suburbs recalls the image of a detached single-family house with at least one person commuting into the central city for work by car. In contrast, cities were largely downtown business areas, dominated by office buildings and government agencies and with limited residential housing options. 

While this delineation may have been clear in the mid-20th century, the lines between cities and suburbs have blurred as the picture of today’s metropolitan areas has grown much more complicated. Yet, the availability of data regarding the definitions of cities versus suburbs has failed to evolve, reinforcing the old dichotomy of city versus suburb and ultimately leading to incomplete if not inaccurate conclusions about our metro areas’ structure and needs. 

For context, in 1950, there were less than 200 standard metropolitan statistical areas (MSAs), most of which contained only one so-called principal city. The Census Bureau defines a principal city as either the largest city in a metro area, or one that meets certain population and employment criteria. As population and employment have continued to grow in many parts of our metro areas, including some previously described as suburbs, so, too, have the number of areas that meet the principal city criteria. By 2017, the number of MSAs had surpassed 300, and for many of those MSAs, multiple principal cities exist. The Washington, D.C., metro area, for example, sprawls four states and has nine principal cities, including one 45 miles from D.C. itself. 

This traditional approach draws rather arbitrary lines around parts of our MSAs, using overly simplistic categorizations to lump together such disparate geographies as Frederick, Md., and downtown Manhattan. The reality is that many geographies possess some combination of both urban and suburban characteristics, which is why we need more refined measures of our urban and suburban areas. 

Some researchers have begun to more clearly delineate suburbs using alternative frameworks. Adding to these works, this Research Notes investigates how certain housing characteristics vary by density level compared to Census-defined city versus suburb status. This approach shows how even accounting for density but still separating cities and suburbs can be limiting. 

Measuring the Housing Stock

One of the big hurdles in making quantitative city-suburb comparisons is the fact that data from the Census Bureau fails to provide data that easily conform to suburban geographies defined by the Office of Management and Budget (OMB). However, principal city definitions are available from Census and can provide some direction, albeit with a few caveats. 

Because some principal cities are located within a county, it is often necessary to use a smaller geography than county delineations look at suburban areas. The Census Bureau does provide files that allow the use of Census tracts, which do conform to “cities” and “suburbs” categorizations. However, recent American Community Survey data are, unfortunately, only available as aggregated tables from the American FactFinder, which means that limited data, mainly summary statistics, are available. While admittedly some of these tables have large margins of error, in general, they provide valuable insight into an otherwise difficult issue to study.

Using this framework, we can begin to examine the composition of the housing stock in both city and suburban areas. The figure below shows a clear disparity in the housing stock between the two. The average multifamily (+2) share of units in city Census tracts (40 percent) is significantly higher than suburban Census tracts (18 percent), and the opposite holds true for the average single-family share. 

Density as the Unit of Measurement

However, as additional geographies are established as principal cities, this analysis of city-suburb contrast may be less useful as it offers a very broad-brush look at increasingly diverse metro areas. To better reflect changing development patterns, density analysis offers a logical alternative to this binary city-suburban classification. 

Using the number of households per square mile as the measure of density, we can identify five major density breakouts. Most Census tracts (53 percent) fall in the lowest density category, and 80 percent of those tracts are in areas defined by the OMB as suburbs. In contrast, 90 percent of the highest density Census tracts are in cities. 

A more nuanced relationship between the share of multifamily and single-family units begins to emerge when density is used as the unit of measurement rather than the Census designation of city or suburb (Figure 2). As density increases, the average multifamily share gradually increases, while the opposite relationship occurs for single-family—as density decreases, the average single-family share gradually increases. 

Moreover, this relationship between density and housing composition holds regardless of whether or not a Census tract is categorized as being located within a city or suburb. As density increases, the average multifamily share increases, while as density decreases, the average single-family share increases. The one arguable exception perhaps being in the lowest-density category, where there is the greatest variance in multifamily’s share of the housing stock (Figure 3). 

Implications for Apartments

These more nuanced shifts in the composition of the housing stock suggest that density can be a more effective tool for distinguishing apartment trends, however it is not possible to test this using publicly available data given the lack of microdata available at the Census tract level. 

To further test the framework, we conducted a comparative analysis using the results from the 2017 NMHC/Kingsley Associates Renter Preferences Report, which provides descriptive information about apartment residents in addition to their preferences on a variety of housing features and amenities. While the survey data are not available at the Census tract level, they are available at the zip code level, which allowed us to analyze the data using density as well as traditional city/suburban designations. We began by examining car ownership trends among apartment renters.

Figure 4 below shows that the distribution of the number of vehicles per household based on city/suburban status is very similar. Using just the city/suburb distinction, one would conclude that the number of vehicles per household does not differ noticeably whether a household is in a city or a suburb, and carless households are a very small part of the overall sample.

However, when density is factored in, a much different picture emerges (see Figure 5 below). For the highest-density category, 39 percent of households had no car. That density category also had the highest use of mass transportation for commuting. While the highest-density category accounts for only about 5 percent of the overall survey respondents, it’s clear that renter households in these areas have a significantly different relationship to car ownership than those in lower density areas. This is a trend that traditional analysis of city/suburban designations was unable to surface. 


Housing types differ greatly between city and suburban areas. Cities have traditionally been higher density areas, with more multifamily units than in areas traditionally known as suburbs. However, as our research shows, the term “city” should not be confused with “high density.” Clear differences in the composition of the housing stock and resident characteristics emerge when viewed through the lens of density that would otherwise go unnoticed when using traditional city or suburban designations. As development continues to change the landscape of our metro areas and more areas qualify as “cities,” density may prove a more reliable indicator of housing trends.

About Research Notes
Published quarterly, Research Notes offers exclusive, in-depth analysis from NMHC's research team on topics of special interest to apartment industry professionals, from the demographics behind apartment demand to effect of changing economic conditions on the multifamily industry.
Questions or comments on Research Notes should be directed to Mark Obrinsky, NMHC's Senior Vice President of Research and Chief Economist, at mobrinsky@nmhc.org or 202/974-2329.
Print-Friendly Version RSS Search Back Issues
Contact Us
1775 Eye Street, N.W., Suite 1100,
Washington, D.C. 20006
202 974 2300 Office
Manage My Email / Unsubscribe
Please add ‘news@nmhc.org’ to your email address book or Safe Sender List. If you are still having problems receiving our communications, see our White List page for more details.