Laura Bliss is CityLab’s west coast bureau chief, covering transportation and technology. She also authors MapLab, a biweekly newsletter about maps (subscribe here). Her work has appeared in the New York Times, The Atlantic, Los Angeles magazine, and beyond.
The surprising results of a new study tracked a decade of real family expenses, and it calls into question a fundamental assumption of affordability research.
In the U.S., whether you own a car is the main determinant of how much you spend to get around—gas alone costs the average American nearly $3,000 annually. So it stands to follow that living in neighborhoods with good public transit access should reduce your transportation expenses. This notion of “location efficiency” is one of the reasons that many housing advocates (including supporters of the now-dead SB 827, the radical transit-oriented housing proposal in California) are so passionate about building affordable units near train and bus stops.
But does moving to a transit-rich neighborhood mean a person’s transportation costs systematically come down? Not according to a new paper that studies a decade’s worth of income dynamics for 11,000 American families, rich and poor. The study, published recently in the journal Housing Policy and Debate, finds almost no relationship between lower transportation spending and neighborhoods with better bus and subway connections when studied through the lens of real-world household spending.
The findings may give pause to those interested in how housing and transportation, the largest household expenses in the U.S., intersect in wallets and bank accounts. Years of research and widely-used data indices from the community development organization Center for Neighborhood Technology and the U.S. Department of Housing and Urban Development have implied that the more transit-rich a neighborhood is, the less people spend on transportation, and therefore the more “affordable” that area is. But while these indices are built mostly on models that estimate transportation expenditures for typical households, this paper studied direct measurements of these expenses of real people over time. The results did not match up. (That said, a leader of the prevailing school of location-efficiency said that there is good reason for this, and called into question the premise of the paper. Read on for his rebuttal.)
Authors Nicholas Klein, a transportation planning professor at Cornell University, and Michael J. Smart, a professor of urban planning at Rutgers, relied on a source of data that had not previously been used for studying this topic: part of the Panel Study of Income Dynamics (PSID). This nationally representative household survey began in 1968 and has tracked in detail the basic expenses of the same thousands of American families since. Klein and Smart used a confidential, geocoded version of PSID that reflected the total transportation expenditures (including vehicle purchases, gasoline, repairs, insurance, transit fares, and others) of 11,000 families across income brackets and neighborhoods, between 2003 and 2013. It was only one set of numbers, Klein and Smart acknowledged, but “it punches above its weight by following the same families over time as they move, and with rich detail on actual expenditures,” they said in an email to CityLab.
The authors followed three methods to test the location efficiency hypothesis against the ten years of survey data. First, they took “snapshots” of the PSID families at different moments in time, examining the basic relationship between lower transportation costs and higher transit access (as measured by the number of jobs accessible by public transit within 30 minutes). They found a weak correlation, not the strong diagonal line they expected.
Next, the authors charted how a change in transit accessibility affected a change in a family’s transportation spending, immediately after moving from one neighborhood to another and then two, four, eight, and ten years after moving. Again, they found almost no systematic effect.
“Some people do move and see their costs go down—but there are almost as many people whose expenses go up,” Klein said in an interview.
Finally, they built a model that looked specifically at how transportation expenses changed, in actual dollars, among families who relocated to better or worse transit environments. Again, across income brackets, they saw only a weak relationship between the two. Only by moving from the very extremes of the transit accessibility did families see more significant savings. “By going from the very bottom of the transit access scale to the absolute top, you’d save $58 per month,” Smart said. But such a transition was not the norm.
Likewise, population density, employment density, walkability, and neighborhood compactness did not significantly affect transportation spending, either, the authors found. While some families spent less on getting around when the shape of their neighborhoods changed, almost as many families spent more. Indeed, the transportation expenses of non-movers and movers alike were largely unaffected by transit access. All of this was contrary to the strong relationships between transit access and household expenses that Klein and Smart expected based on the existing literature.
The simplest way to think about these findings is in terms of car ownership, the big-ticket item among household transportation expenses. On paper, it’s easy for planners and policymakers to believe that families who live in neighborhoods where they could give up their cars do, in fact, give them up. But according to this study, transit access turns out to be low on the list of factors that actually push people to give up their cars. Much more important are the number of adults and jobs in their households, how many children they have, their annual income, and the cities they moved to and from.
“It’s not that transit doesn’t help people,” said Klein. “But lots of people who are making totally rational decisions not to use transit and to have their car are doing it to help their family.”
The implications for housing planners and decision-makers, then, is that policies that assume good transit will lead to lower household spending won’t necessarily work for all families, especially those who aren’t already intent on giving up their cars. That includes “location efficiency mortgages,” or LEM, a transit-oriented housing financing idea that was piloted in the 1990s and early 2000s, and which petered out around the recession.
The paper has inspired conversation among housing and transportation researchers, who called attention to the new study on Twitter last week. Dan Immergluck, a professor at the Urban Studies Institute at Georgia State University, tweeted that the paper “calls into serious question” the focus among housing advocates on building affordable units only near transit. There may be other good reasons to do so, he added in an email to CityLab, but “locating affordable housing near transit may not lead to as much cost savings as such modeled estimates imply.”
Lisa Schweitzer, a professor of public policy and transportation planning at the University of Southern California, tweeted, “I have always [had] trouble with the basic idea of the LEM.” Just because someone has “access” to an amenity, she continued, doesn’t mean they’ll use it.
The findings of this paper may call to mind a paradigm shift in the realm of affordability research. For more than a decade, affordability indices from CNT and HUD have been widely used by researchers, planners, policymakers, and nonprofits to study and guide real-world development and policy decisions. They also encourage prospective homeowners to use them to evaluate where to live. Both indices factor in transportation in addition to housing costs as measures of a neighborhood’s affordability at the census block level. They rely on the assumption that the built environment is a key determinant of how much a person spends on transportation; this paper suggests that this is not the case.
However, this is only one paper, and no single study is likely to settle any academic debate. More research could be used to validate Klein and Smart’s findings. Furthermore, Scott Bernstein, the president and co-founder of CNT and one of the principal authors behind its widely used Housing + Transportation Affordability Index, took issue with the idea that this paper poses a serious challenge to his own years of research. In fact, he felt that the paper’s premise mischaracterizes the fundamental intentions of his work.
In a detailed statement provided to CityLab, Bernstein said that the CNT index uses a model drawn from peer-reviewed research and that includes many factors that drive household transportation costs. It has been tested against “real” data, including vehicle odometer readings from millions of households. Klein and Smart used overly narrow samples and metrics from which to draw broad conclusions, Bernstein argues. And fundamentally, while Klein and Smart are measuring the expenses of individual families who move, CNT’s index is supposed to measure the potential affordability of a place. “People can, and often do, spend less in areas with high location efficiency,” Bernstein said.
But Klein and Smart said that studying families who change neighborhoods gets to the heart of the affordability question. It is individuals, after all, who bear the costs of transportation, not neighborhoods. If a place is supposed to be more affordable because of its good transit access, people who move to that place should expect to see some savings, they said.
Perhaps the paper’s significance was best summed up by Smart’s mother, who was apparently unfazed by its somewhat surprising conclusion. “She was like, ‘Of course,’” Smart recalled. “’Everyone loves cars. It doesn’t matter where you live.’”