Location patterns of service activities in large metropolitan areas: the Case of São Paulo

We present a set of detailed evidence about the location patterns of service activities in the largest and most important Brazilian metropolitan region, the São Paulo Metropolitan Region (SPMR). Different from previous analysis of this big urban agglomeration, our results are obtained using a unique dataset of geocoded firms and a distance-based measure of firms’ location, thus not susceptible to the modifiable areal unit problem (MAUP). We find that around 89% of 3-digit service sectors present significant defined location patterns and, based on maximum distances where significant location patterns are observed, identify spatial location of clusters of some activities. Our results also indicate that firms’ activities of FIRE (finance and real estate), IT-related services, and high human capital-based services present the highest probability of location at shorter distance from each other. The tendency for location at shorter distances between firms engaged in these activities contrasts with the more decentralized patterns observed for firms involved in retail and urban infrastructure services. Additional results indicate that both the location patterns of activities and the degree of proximity or agglomeration of firms are positively associated with human capital, the degree of product differentiation, and the degree of inter-sector dependence between activities.