Andrea Chandrasekher of UC-Davis has a history of writing important papers on important subjects. She has done so again, with The “Good” Starbucks: Consumer Redlining in Large American Chain Stores. Here’s the abstract:
Racial discrimination in the retail realm has been well-documented in the academic literature. However, past studies have focused on retail redlining, a discriminatory practice whereby retailers avoid opening stores in predominately Black neighborhoods, leaving consumers without a viable place to shop that is close to their homes. This paper focuses on a different, understudied type of retail discrimination called consumer redlining. This type of discrimination occurs when chain stores in Black neighborhoods offer a lower-quality consumer experience than those same chain stores in white neighborhoods: shelves are understocked, food is sold past its expiration date, and stores are understaffed leading to longer wait times, dirtier facilities and bathrooms, and disorganized inventory. In this paper, I document and then explore the causes of consumer redlining in over 400 American chain stores and eight large metropolitan areas. Using a unique data set of Yelp ratings and reviews, I estimate a difference-in-differences specification with chain store fixed effects to capture the within-chain consumer experience gap between Black-located and white-located stores. I supplement the Yelp data with a wide set of demographic controls from the American Community Survey, neighborhood market potential controls from ESRI, property and violent crime controls and franchise status controls. Holding constant these factors, the results show that for chain stores that are located within both Black and white neighborhoods, the Black-located stores have, on average, a one-quarter star lower Yelp rating than the white-located stores. The result is most prominent in the food industry (fast-food and coffee/snack shops) and big box stores (particularly discount stores) and is robust to the inclusion of over 238,000 Yelp user fixed effects. Using an event study, I also show that this result is not likely to be driven by any potential racial bias on the part of Yelp users. Finally, the paper concludes by exploring potential causes for the persistence of this racial consumer experience gap in equilibrium.