Retail Prices in a City

We study grocery price differentials across neighborhoods in a large metropolitan area (the city of Jerusalem, Israel). Prices in commercial areas are persistently lower than in residential neighborhoods. We also observe substantial price variation within residential neighborhoods: retailers that operate in peripheral, non-affluent neighborhoods charge some of the highest prices in the city. Using CPI data on prices and neighborhood-level credit card data on expenditure patterns, we estimate a model in which households choose where to shop and how many units of a composite good to purchase. The data and the estimates are consistent with very strong spatial segmentation. Combined with a pricing equation, the demand estimates are used to simulate interventions aimed at reducing the cost of grocery shopping. We calculate the impact on the prices charged in each neighborhood and on the expected price paid by its residents - a weighted average of the prices paid at each destination, with the weights being the probabilities of shopping at each destination. Focusing on prices alone provides an incomplete picture and may even be misleading. Specifically, we find that interventions that make the commercial areas more attractive and accessible yield only minor price reductions, yet expected prices decrease in a pronounced fashion. The benefits are particularly strong for residents of the peripheral, non-affluent neighborhoods.