[memph]

I'm hoping that many of these losses were just due to the housing crash and will be reversed with the recovery + shift back towards cities. If you have a 2-3 percentage increase in the share of housing units that are vacant in 2010 vs 2000, plus a decrease in household sizes over a large area, that can add up.

In the case of Chicago though, it's not just the South Side, there's been relatively significant losses in the north side as well. The % loss in the North side is smaller, but there's a lot of people living there so it still adds up. There's been losses in the Hispanic areas too, not sure if that's been due to reductions in household sizes or abandonment.

Anyways, I'm working on the next set of criteria now, which will add census tracts where the non-auto mode share and density is about the MSA average (weighted density for density) to the tracts that qualify based on 1960 densities. That should expand the areas included in the cores of sunbelt cities by multiples. In older cities where the core that's based on 1960 densities is already large, it should change little - if at all (already did Buffalo, not a single census tract was added).