There was a study I saw as an undergrad that used all the standard factors and also used zip code -- and that worked surprisingly well to explain a lot of the differences. But using race rather than zip code, actually was still better -- and of course zip code may well have simply been a race proxy.

I think zip code though will likely be more satisfying to people. Although the conservative wing of the supreme court has already said that they don't like the top 10% rules for admission, so I suspect they'll also not be kind to zip code since the idea is similar in spirit.