Residential self-selection and the reverse causation hypothesis