Spatial relationship between sociodemographic and retail access data and e-commerce deliveries: the case of Belo Horizonte (Brazil)

Many negative externalities are associated with home deliveries, the main ecommerce delivery destination. Despite many solutions that address this problem, the lack of understanding of the spatial pattern of urban deliveries makes it challenging to implement these strategies. This paper analyzed the spatial relationship between sociodemographic, retail and e-commerce deliveries in Belo Horizonte (Brazil) by using official data at the neighborhood level (number of retail shops, gender, income, age, race, and household size), and e-commerce deliveries performed by a transportation company. Global Moran's I indicated the spatial dependence of the e-commerce deliveries. Results of a geographically weighted regression model showed a positive spatial effect of retail, women, Asian population, age from 20 to 29 years old, and income. In addition, a negative spatial effect was identified for the size of the household, 18 to 19 years old, and the black population. Furthermore, the estimated coefficients show small spatial variability, indicating homogeneity in the spatial relation. The uniformity of the parameters indicates that alternative strategies can be implemented throughout the territory to reduce e-commerce deliveries.