In 2006, the Brookings Institution worked with the Center for Transit Oriented Development and the Center for Neighborhood Technology to study the transportation patterns of the U.S.’s low-income population. Until then, many researchers and policymakers had assumed that larger and wealthier households owned more vehicles—and more expensive ones—and drove more miles overall. But the 2006 study found that transportation methods had less to do with household income and more to do the neighborhoods in which those households were. The researchers concluded:

Even among wealthy households, neighborhood characteristics influence how much is spent on transportation and how many vehicles are owned, given that the characteristics of place also shape transportation demand. Neighborhood characteristics such as density; walkability; the availability and quality of transit service; convenient access to amenities such as grocery stores, dry cleaners, day care, and movie theaters; and the number of accessible jobs shape how residents get around, where they go, and how much they ultimately spend on transportation.

To regular readers of this site, these conclusions are chock full of “duh.” Of course transportation costs are highly dependent on where one lives. But transportation costs are not always considered in discussions of affordable housing. As defined by the Department of Housing and Urban Development, affordable housing costs less than 30 percent of a household’s income. But when necessary transportation costs—gas prices, car maintenance, monthly transit passes, etc.—are added into the mix, those percentages go way, way up.

Which parts of the country host the most actual affordable housing? That’s the question behind new work from Shima Hamidi, an assistant professor of urban planning at the University of Texas at Arlington, and Reid Ewing, professor and director of the Metropolitan Research Center at the University of Utah’s College of Architecture and Planning. Unlike the transportation affordability tool developed in that 2006 study, or a similar tool offered by HUD, Hamidi and Ewing’s new index uses disaggregate, household-by-household data, rather than aggregate data based on census tracts or blocks.