And let's be clear: those fields exist. I didn't get into the Peter Stark program at USC because my last name isn't Coppola. At the same time, my first week on the job, I had my coffee made by a guy who had just graduated from the Peter Stark program at USC. Saved me $200k right there and then, I did. If you're going to work for Bain or Goldman Sachs or their ilk, a Harvard degree probably matters. If you're going into research, studying under someone who knows their shit matters. Dated a girl whose sister had a theorem named after her by the time she was 20; her grad school choices were MIT, Stanford and Columbia. MIT gave her a $4k a year scholarship, Columbia wanted full sticker, and Stanford offered her a $20k a year stipend. So she went to Stanford and is now at the Max Planck Institute.
Grad school is a whole 'nuther beast. The advice I would give out about grad school (in the sciences, because that's all I know) is (a) you don't need to go to an elite university for undergrad to go to an elite university for grad, and (b) don't pay attention to schools when deciding where you want to go--pay attention to individual researchers. There are far higher concentrations of good quality researchers at elite universities, and that's pretty much what makes them elite. Firstly, those people aren't teaching your intro to chemistry course. Second, there are tons of great researchers at state schools, they just may be less densely concentrated. Prospective grad students should decide what field they're interested in, find the best profs in that field, then contact the prof directly. Then go to that school if and only if they offer you free tuition plus stipend. No one should ever pay for science education (at the PhD level). Ever.