||In Latin American cities there is a high correlation between the location chosen by poor households and their income level; however, it is difficult to identify to what extent they live there by choice because it maximizes the returns to their efforts or by restrictions that pull them to locations that make them poorer. We define the former case as unrestricted sorting in the urban economics context, while the latter is assumed to be the commonly used definition of segregation. The main hypothesis is that segregation exists when these returns vary across space while they should not. That is when households cannot profit equally across space even if they have comparable characteristics. Results show that segregation, as opposed to individual characteristics, explains one fourth to one third of the mean income difference between locations in Bogota, Colombia.