New Yorkers are more likely to choose to eat out at a restaurant in a neighborhood with demographics that reflect their own individual race or ethnicity, find Columbia’s Donald Davis, Chicago Booth’s Jonathan Dingel, Pompeu Fabra’s Joan Monràs, and Princeton’s Eduardo Morales.
Their research illustrates that although it has been illegal to discriminate on the basis of race in the United States for more than a half century, some aspects of American life, such as eating out, remain fragmented along racial lines.
In the 1950s and ’60s, protestors pushed to desegregate spaces such as buses and lunch counters. The Civil Rights Act, passed in 1964, prohibits discrimination on the basis of race or ethnicity in public accommodations. But in terms of consumption venues such as restaurants, how racially integrated are we?
The constraints of traditional data sources have made this question difficult to tackle. Public consumption surveys do not typically capture business and location data along racial lines. But Davis, Dingel, Monràs, and Morales used the online review platform Yelp to analyze dining choices of New York City residents, culling information about home and work locations and racial identities from profiles and more than 18,000 restaurant reviews between 2005 and 2011—examples of the kind of “digital exhaust” often generated by smart devices and online platforms.
They find that while consumption choices tend to be less segregated than neighborhoods themselves, it’s still possible to use the racial composition of a neighborhood to predict the race of consumers most likely to patronize its restaurants. Controlling for the time it takes to travel to eat, the researchers analyzed the choice of a diner opting between two restaurants that were identical except for the neighborhood: one was in an area that was more demographically similar to the diner’s residential neighborhood by one standard deviation. An Asian diner was 25 percent more likely to choose the restaurant in the more-similar area, they estimate, and a Black diner was 51 percent more likely to do so. For white and Hispanic diners, whom the researchers didn’t distinguish between when classifying Yelp users’ profile photos, the estimate was 31 percent.
When the researchers compared the trade-off between demographic difference and transit time, they again observe the same behavior. For an Asian diner to choose a restaurant that was in a more demographically different neighborhood by one standard deviation, the restaurant had to be 21 percent closer in terms of travel time. For a Black diner, the calculation was 44 percent.
Transit time may play some small role, Dingel says, but much of the main finding is due to social frictions, the fact that demographic differences tend to predict restaurant-visit decisions. These may result from friendship networks if people frequent restaurants near friends and family who have similar demographics. The results could also reflect information frictions: someone in a white neighborhood might not be as aware of their dining options in Black neighborhoods, although Dingel points out that information about all restaurants are easily available on Yelp.
One result of the continuing segregation is that people miss out on having shared experiences in spaces such as restaurants, where casual contacts could help break down stereotypes. Also, in gentrifying areas, existing residents may find themselves with fewer places to shop or eat out. When the researchers computed the change in welfare that Harlem residents would experience if the area were to become more like the whiter, richer Upper East Side, they find a significant decrease in predicted patronage of restaurants on the basis of the associated demographic changes.
“The segregation of consumption that we document underlines the role of race and ethnicity in our everyday lives. Lack of interactions among diverse consumers in consumption venues is both a cause and a consequence of the fragmentation of American life along racial and ethnic lines,” Dingel says.