Shocks to the overall retail rent level create vacancy at the edge and eventually lead to other uses of the space at the edge. We verify both empirically. (4/4)

This structure can be explained by walking behaviour of shoppers and leads to different slopes of the rent gradient for different types of shopping districts. Longer stays imply flatter gradient, for example. (3/4)

We study the spatial structure of the Dutch shopping areas using detailed data on shop-level rents, vacancy and transformation to other uses. We show that rents are highest at centre and decrease outwards until other land use yields higher rent. (2/4)