Though large-scale suburbanization remains one of the most striking features of post-World War II America, there remains considerable debate about the economic forces underlying it.
This paper uses a new methodology to untangle common explanations for suburbanization and isolate the specific driving forces behind two observed trends: Americans’ greater willingness to commute, and the shift in location of workplaces and residences to lower-density areas over time.
It demonstrates that Americans’ increased willingness to commute was largely driven by factors that eased commuting, for example the expansion of the interstate highway network and the reduced real cost of car ownership. The author’s analysis shows that if “commuting frictions” in 2000 were what they were in 1970, a much smaller share of people would live in the suburbs, even holding constant the attractiveness of their workplace and place of residence.
In contrast, changes in workplace or residence attractiveness do help explain the shift towards lower densities from 1970 to 2000. The need for larger workplaces (for manufacturing or other purposes) or better schools helps explain why jobs and people moved away from the city, regardless of individuals’ willingness to commute.
To untangle these driving forces, the author first provides reduced-form evidence on observed trends in suburbanization over time. They then use a type of spatial model, characterized by a structural gravity for commuting, to discriminate between leading explanations for these trends.
To build their model, the author uses U.S. Census Bureau data on the number of bilateral commuters from each workplace county to each residence county (in the 48 contiguous U.S. states plus Washington DC) for every census decade from 1970 to 2000. They combine these data with information of the geographical characteristics of counties, including geographical land area and the latitude and longitude of each county’s centroids.
Observed changes in suburbanization and commuting
The author’s reduced-form analysis reveals three large-scale changes in commuting patterns from 1970 to 2000:
- People became more open to commuting to work. Between 1970 and 2000, the median share of residents who work in the county where they live fell from 87 to 71 percent. Counties for which less than 50 percent of their residents work in the county where they live increased from around 5 percent to about 18 percent.
- Workplaces and people moved away from major cities. There was a dispersion of both employment and residents from central cities, with the distributions of the share of employed people who work in each county and the share of employed people who live in each county moving toward areas with lower population densities.
- The shift in spatial concentration of workplaces was greater than the shift in spatial concentration of where people lived. While the location of workplaces was more spatially concentrated than the location of residences in both 1970 and 2000, workplaces displayed a greater shift towards decentralization over time.
Results from the spatial model
The author’s spatial model shows that changes in two-way commuting frictions are the dominant force behind the author’s first observed change: Americans’ willingness to commute. Holding workplace and residence attractiveness constant at their 2000 values and changing only bilateral commuting frictions to their 1970 values, the median share of residents who work in the same county where they live rises from 71 to 85 percent, almost as large as the rise to 87 percent observed in the data.
In contrast, changes in residence attractiveness and workplace attractiveness are more important in explaining the second and third observed changes: The shift in workplaces and residences toward lower-density areas over time.