From stores to websites: the spatial effects of e-commerce logistics on brick-and-mortar retail

Abstract

This thesis studies how the expansion of e-commerce-related logistics infrastructure affects local brick-and-mortar retail markets in Mexico. Using establishment-level data from DENUE for 2015–2025, I construct an annual panel of physical retail establishments at the Área Geoestadística Básica (AGEB, or Basic Geostatistical Area) level. Treatment is defined as exposure to at least one logistics center within a 20 kilometer radius, and identification comes from the staggered rollout of these facilities across space and time. I estimate two-way fixed effects models as a benchmark and rely primarily on staggered differencein-differences estimators proposed by Sun and Abraham (2021) and Borusyak et al. (2024). The results show that exposure to nearby logistics infrastructure reduces the number of local brick-and-mortar retail establishments by approximately 7 percent in the study period. This effect is stronger in AGEBs with higher baseline retail clustering, consistent with a spatial competition mechanism in which e-commerce reduces the value of physical proximity and generates stronger pressure in less differentiated local markets. To evaluate Hypothesis 3, I also examine whether exposure is associated with changes in the spatial organization of observed stores. The estimates suggest greater average distances and more regular spacing between establishments after exposure. However, the evidence supporting parallel trends for these spatial outcomes is mixed, limiting their causal interpretation. Overall, the main results indicate that e-commerce logistics affects the size of local retail networks, with effects that vary according to their initial spatial composition.