Income-specific mobility networks reveal persistent socioeconomic representations of metropolitan functional structure

Metropolitan functional structure is commonly inferred from population-wide mobility flows, but the extent to which this representation is shared by different socioeconomic groups remains unclear. We address this question using harmonized Origin-Destination household travel surveys for the São Paulo Metropolitan Region in 2007, 2017 and 2023. Weighted directed networks were constructed for five income quintiles and the aggregate population and evaluated using complementary network indicators, permutation tests, bootstrap intervals and equal-person rarefaction. Lower-income networks were consistently sparser, more compartmentalized and characterized by lower duration-based reachability, whereas higher-income networks contained more direct interzonal connections and weaker community separation. The functional-center sets of the poorest and richest quintiles overlapped by less than 10%, and none of their ten highest-PageRank zones coincided in 2023. These differences were persistent but not static: community compartmentalization was the most durable socioeconomic distinction, while connectivity gaps narrowed and centrality concentration changed more substantially. Aggregate networks also represented income groups unevenly, exhibiting consistently lower center-set similarity with the poorest quintile than with upper-income groups. The findings reveal a socially differentiated polycentricity in which multiple centers exist but do not constitute the same functional metropolitan system for all populations. Income-disaggregated networks therefore complement aggregate representations and provide a planning-relevant basis for evaluating metropolitan integration beyond population-wide averages.