We tend to think of urban sprawl in America as a product of the Interstate Highway System built in the 1950s and 1960s. Metro area residents who might have been inclined to live near work in the city took the chance to head up the road, find a parcel of land for a single-family home, and commute into work by car. Others followed and pushed development farther out until we got the sprawled out metros we know today.
Some new work published today in the journal PNAS challenges this timeline—showing evidence of sprawl dating back to the 1920s. Using precise, street-level data at the county level, Christopher Barrington-Leigh of McGill University and Adam Millard-Ball of UC-Santa Cruz report that sprawl was rising well before 1950, then grew steadily through the 1990s. The researchers also conclude that U.S. sprawl peaked around 1994 and has been falling ever since.
“Perhaps our most interesting result was that new streets, and the street network being built today, is much less sprawling than in the past,” says Millard-Ball. “I think the policy is very encouraging; perhaps some of the recent policy efforts and planning efforts have been paying off.”
The rise and fall of sprawl
For the study, Barrington-Leigh and Millard-Ball viewed sprawl as a poorly connected street—a dead-end, for instance—that stymies access to outside areas and promotes driving. That definition has some advantages over others, they argue, specifically density. Street connectivity is easier to track over time and more closely connected with vehicle travel and emissions, the very products of sprawl that policymakers and planners hope to alleviate.
Using land parcel records from a sample of U.S. counties, the researchers calculated street connectivity several ways. They looked closely at street nodes, or the number of connections a street makes with another; with this measure, a typical intersection (with four nodes) scores as less sprawling than a cul-de-sac (with one). They studied an area’s node average as well as its proportion of four-plus-node street and of dead ends.
This digging showed Barrington-Leigh and Millard-Ball that sprawl started to rise in the mid-1920s—indicated by a drop in average street node scores. The trend suggests that developers started to copy the early cul-de-sac planning that occurred in the likes of Radburn, New Jersey, in 1928. When the interstate system emerged, says Millard-Ball, it perpetuated a disconnect street trajectory many metro areas were already following.
“Certainly street-level sprawl predated the era of mass car ownership,” he says. “The interstates, if they had an effect, it would be more likely through their effect on car-dependence—the incentive for developers to build neighborhoods which are car-oriented.”
Then sprawl construction started to level off in the 1990s, according to the street connectivity data. Mean nodes in the study sample rose from 2.6 in 1994 to a little over 2.8 in 2012—a decline of about 9 percent. While individual metro areas differ in their means, the trends tend to look the same from place to place; the chart above shows similar patterns in Miami, New York, Los Angeles, and Seattle.
Millard-Ball recognizes some limitations of measuring sprawl by street connectivity rather than by density or distance from a city center. For one thing, making a four-way street arguably creates more opportunities to drive than a three-way street—thus elevating total vehicle miles and emissions. But all else being equal, he says, it’s better for metro areas to connect their streets than to isolate them, especially since street patterns are so difficult to change over time.
“Any planner who’s tried to build even a pedestrian path that connects up to even a dead end will know the big backlash that attracts,” he says. “Dead ends are a good way to transit-proof, density-proof your development for a long time into the future.”
What changed in 1994?
The Barrington-Leigh and Millard-Ball can’t point to any causes of what happened in 1994 to shift street sprawl trends. But they have some ideas. It’s around this time, they argue, that smart growth policies began to spread more rapidly throughout the planning community; Congress for the New Urbanism, for instance, was founded in 1993. Around this time, or shortly after, many progressive local governments started to discourage cul-de-sac development and push for connected grids.
“It’s a very gradual turnaround,” says Millard-Ball. “It’s not that suddenly people woke up in 1994 and decided, ‘Oh, we’re going to build grids only from now on.’” He also stresses that policies focusing on density likely played a role in the transformation, especially since the two often go hand-in-hand.
You can see some evidence for this theory in the case of metropolitan Dallas. The city’s 1998 comprehensive plan says residential neighborhoods “shall be served by a grid street system, which minimizes the use of cul-de-sacs, double-frontage lots and walled subdivisions.” Dallas-Fort Worth ranked second in the metro area street connectivity ratings gathered by Barrington-Leigh and Millard-Ball, with an average node score of 2.81 from 1991 to 2013 (against a national average of 2.6).
Barrington-Leigh and Millard-Ball found even more evidence that planning trends seem to be shifting sprawling street construction at the county level. They tracked the biggest changes in node average from the 1993-1997 to the 2008-2012 periods, and found a top three of Travis (Texas), Mecklenburg (North Carolina), and Alachua (Florida). These counties are home to Austin, Charlotte, and Gainesville—three cities with strong recent connected streets policies.
(Researchers and other interested readers can check out the whole data samples, the full metro and county rankings, and animations of the density changes here.)
Again, the suggestion here isn’t that street connectivity is a perfect proxy for sprawl—or, conversely, that grid-style layouts eliminate car reliance. Take the example of Stapleton, Colorado, a neighborhood just outside of Denver designed to embody New Urbanist principles. While Stapleton gets highlighted in the Barrington-Leigh and Millard-Ball paper as an area with high nodes, it’s been criticized for having wide, “vehicle-oriented” roads and excess parking.
What the figures show more than anything is that sprawl is persistent. “So areas that were sprawling in the past seem to be continuing to sprawl more today,” says Millard-Ball. “It’s more difficult to build a walkable neighborhood if there’s nowhere to walk to.” On the flip-side, that also means the progressive, grid-like development that seems to have taken hold in the mid-1990s should direct the character of cities moving forward. The researchers conclude:
Just as the existing stock of locked-in sprawl from the mid- to late 20th century represents an enormous inertia, newly developed, connected street patterns will continue to affect vehicle travel and emissions for the next century and beyond.