The strongest evidence yet that parking spaces cause more driving
New research offers an important lesson on how urban design shapes people’s behaviors.
Cities typically require developers to create a certain amount of parking for every new housing unit in a project. Space by space, this policy takes a toll on affordability. Parking is very expensive to build, so developers bundle these parking costs into the housing costs for all residents, whether or not households even want a parking space at all.
Some cities have finally started to move away from parking requirements, with Berkeley and Toronto being the most recent North American examples. But the policy remains entrenched in most local rulebooks — and in local minds. Many neighborhood groups will oppose new development projects that don’t require parking, out of fear that newcomers will come for their own spaces.
The assumption here, behind both the policy and the pushback, is that people will always choose to drive, and the urban environment should be designed to accommodate that inevitable choice. But new research shows how that assumption is often backwards — offering the strongest evidence yet that parking doesn’t just follow driving in cities, but can actually cause it.
The new work comes from a group of urban planning scholars at UCLA and UC-Santa Cruz, led by Adam Millard-Ball, and is set for publication in an upcoming issue of the journal Urban Studies. Using an innovative and elegant study method, the researchers show clearly that “increased parking causes more car ownership and more driving while reducing transit use.” They continue:
In summary, the evidence from our study robustly supports that urban residents’ transportation behavior — but not their employment — is affected by local features of the built environment, and particularly so by parking.
The conclusion underscores the importance of urban design in shaping behavior. More importantly, the study helps make the case against parking requirements. A move in that direction would not only reduce housing costs but also the pollution and traffic congestion created when cities are designed to prioritize driving over all other travel modes.
What they did
Researchers have tried to show before that parking causes driving, often going so far as to borrow the approaches used by epidemiologists in showing that smoking causes cancer. But the causal connection between parking and driving is a very hard one to draw for several reasons.
The biggest reason is that most people don’t live in random places. Instead, they choose a home for many reasons, including things like available parking (if they prefer to drive) or proximity to public transportation (if they prefer to ride). Researchers can try to control for this self-selection bias, but this type of statistical analysis doesn’t allow for a causal conclusion.
The gold standard for showing causation is through a randomized trial. That would mean placing people into homes randomly and then studying how they live, which Institutional Review Boards frown upon for obvious — and good! — reasons.
This new study distinguishes itself by finding a way to effectively (and ethically) randomize a population: San Francisco’s housing lottery. In San Francisco, inclusionary zoning regulations typically require new developments with 10 or more residential units to provide affordable housing, which is offered to income-eligible households through a lottery. Given the city’s exorbitant housing costs, these lotteries are highly competitive, with the odds of winning a unit around 1.5 percent.
Facing such low odds, eligible households tend to apply for as many affordable units as they can without worrying about, say, whether or not there’s a parking space available. “In essence,” write the researchers, “San Francisco’s housing lotteries provide as-good-as-random assignment of people into homes.”
What they found
In spring 2019 — pre-pandemic — the researchers mailed a travel behavior survey to housing lottery winners in 197 development projects across San Francisco. The short questionnaire, provided in four different languages, asked about typical travel mode (car, transit, bike, walking), car-ownership status, and employment status. Roughly 780 households responded.
When the researchers matched travel behavior to parking requirements, they found “a clear and substantive trend:” as parking supply rose, so did car-ownership. In buildings without any parking, only 38 percent of respondents owned a car. Car-ownership climbed as parking requirements increased, reaching 81 percent of respondents in buildings that required one parking space per housing unit.
Owning a car isn’t the same as using it, but further analysis found a statistically significant relationship between parking supply and driving, too. Generally speaking, households that lived near public transit, or that had good walking or cycling access, tended to use those options more often than households that did not. But when it came to using transit, in particular, the effect of a building’s parking ratio was “more than twice as large” as that of its transit access.
In other words, even in buildings with transit access, parking supply was the stronger pull — increasing driving behavior by the same amount it reduced transit use. When buildings provide free or cheap parking, residents buy a car and drive. But when buildings have transit access without easy parking, residents use other ways to get around.
“Where streets are relatively walkable and transit service is frequent,” writes the research team, “parking emerges as the key factor shaping household travel behavior.”
One final, critical result: the researchers found no connection at all between parking supply and full-time employment status. That’s very important, because it suggests that reducing or eliminating parking spaces won’t negatively impact a household’s ability to keep a job, as is often feared.
What it means
The study represents a significant step forward for urban mobility policy, but it’s not without caveats. For starters, self-report travel surveys are subject to memory errors, so it would be better to use objective trip data in future work. The research took place in only one city, and San Francisco is a fairly large city with a fairly strong transit network. It’s unclear — and perhaps even unlikely — that the results would hold up in places with weaker transit access.
Along those same lines, transit commutes tend to take longer than car commutes in U.S. cities. So even though parking supply doesn’t seem to impact job status, it almost certainly impacts commute time, and thus quality of life. It’s also important to replicate the work in low-income or otherwise underserved communities, as the eligibility thresholds for housing lotteries in San Francisco are relatively high by national standards (up to $118,000 for two people, according to the researchers).
With all that in mind, the findings open the door wide to positive change. Cities can cut back or even eliminate parking requirements, improving housing costs and air quality without harming job access. Cities can also charge residents for residential parking permits, ideally redistributing the revenue into improving the public transportation system.
Technology can accelerate these shifts. Low-cost sensors and parking management software can track spaces in real time, enabling cities or developers to meet local parking needs through shared parking networks. These tools can also make the case for converting excess parking into other uses, such as last-mile logistics hubs, loading zones, green space, even more housing. These parking futures might give less cause for driving, but they’d give great cause for cheer.
Follow Sidewalk Labs with our weekly newsletter or subscribe to our podcast, “City of the Future.”
