Laura Bliss is a staff writer at CityLab, 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.
Research based on years of Yelp reviews finds new grocery stores and coffee shops are indeed indicators of changing populations and rising home prices.
You know it when you see it, or perhaps smell it. Gentrification is that new dog park. It’s the Starbucks on the corner, the yoga studio, and the gradual rise in police presence.
But it’s surprisingly hard to track the exact moment when a critical mass of more affluent people move into a neighborhood and tip property values up—the simplest, if not the most universally agreed upon, definition of the “G” word. Traditional public data sources can fail to pick up the rapid transformation that can occur in a community, since their records are usually updated on multi-year cycles. And government registries usually catalogue businesses in broad categories—you’re not going to find artisanal donut parlors or motorcycle lifestyle shops grouped together by the Census Bureau.
But there’s more than one way to skin a cat café. In a new working paper published by the National Bureau of Economic Research, the Harvard economists Edward Glaeser, Michael Luca, and Hyunjin Kim show how Yelp data can be used to quantify and track neighborhood change, putting a hard spine on what can otherwise be a soft science. Matching up a massive trove of business and service listings from the uber-popular reviews site against changes in housing prices and demographics, they found that Yelp appears to work as a real-time forecaster of neighborhood change. You just have to look at the right types of listings.
Glaeser, Luca, and Kim conducted their study in a few ways. First, in testing a popular theory about signs of the gentry’s arrival, they pulled out all the Starbucks listings on Yelp across the United States dating back to 2007. Combining that information with Federal Housing Finance Agency data by zip code, they found that the arrival of every new Starbucks into a given area was associated with a 0.5 percent rise in local housing prices. Coffee shops of all kinds—artisanal and chain—had a similar relationship.
More broadly, they found that housing prices grew in tandem with the entry of new restaurants, bars, hair salons, convenience stores, and supermarkets. Counting reviews, the Yelp data also captured commercial activity at those businesses, which turned out to be a predictor of rising home values, too.
Who was driving up home prices, and in connection to exactly what type of change? The researchers turned to Census data to glean how demographic changes in New York City compared to a range of Yelp listing types. They focused on three proxies for the latté-drinking class: education level (which is generally correlated with income and housing cost), age (since gentrification is often associated with a change in age distribution), and race (since gentrification is widely perceived as a process by which white people displace people of color).
Fascinatingly, different listing types were more correlated with different demographics than others as they increased within Big Apple neighborhoods. Grocery stores were more strongly associated with demographics than any other listing type—the greater the change in grocery stores in a neighborhood, the greater the change in college-educated white people ages 25-34, the researchers found. “These results seem compatible with the literature on ‘food deserts’ that documents how poorer people live in areas with fewer options for healthy food,” they wrote. Laundromats and restaurants were also strongly associated with college-educated young people, but not very strongly with race. Cafés and bars were strongly correlated with a gain in the college-educated population, but less so with race or age. Overall, race was much more weakly associated with changes in Yelp listings than education or age.
The researchers studied several other listing types—florists, wine bars, vegetarian restaurants—and found some correlations, but none as strong as these. And there were some interesting exceptions in the data. Although Los Angeles, Chicago, San Francisco, and Washington, D.C., showed similar relationships between listings and demographics, only in New York City were laundromats so strongly tied to a gentrifying group. And not all restaurant openings were tied to an uptick in well-educated newcomers: Chinese restaurants showed no such relationship.
This work tags onto a growing body of research and online tools designed to track the spread of gentrification, one of the most contentious issues facing cities today. More and more of them are using novel data sources: Harvard and MIT researchers are taking to Google Street View to track urban change en media res. CUNY scholars have studied business reviews on Yelp to understand how the racial makeup of neighborhoods in transition gets telegraphed through positive or negative customer ratings. And real estate listings from Trulia and Zillow are giving researchers more granular insight into home price changes than traditional data sources.
Still poorly understood, however, is which comes first in gentrifying neighborhoods: the wealthier residents or the “nice” amenities. Also hazy is how the use of online tools like Yelp, Google Maps, and Zillow may contribute to neighborhood change itself.
The aforementioned CUNY study, authored by sociologists Sharon Zukin, Scarlett Lindeman and Laurie Hurson in 2015, pointed at this possibility. “Intentionally or not, Yelp restaurant reviewers may encourage, confirm, or even accelerate processes of gentrification by signaling that a locality is good for people who share their tastes,” they wrote. Beyond something you see and smell, gentrification might also be something you search, map, and review.