Unless you live in a high COL area where even NSF grant money isn’t enough.

The NSF grant money wasn't paid directly to students. The departments still set the pay scale, and had to deal with COL disparities like all employers do. And if your prof couldn't support you, then you didn't work for that prof, or you got a teaching assistantship.

At least in my field, physics, lack of sufficient funding meant you didn't get students.

This whole business wasn't risk free. There were students who dropped out because their prof lost funding. In my view, the risk of not finishing at all is a bigger cost of graduate education than the pay scale.

Lol.. UCLA pays grad students 18k ish. Barely enough to survive. Yes not finishing is a real risk but it’s not true that students can generally survive on that without loans or savings. In other lower cost areas, they can. MIT, Columbia and other schools also have better financial packages / housing options.

I remember even in the 1980s, there were schools where the pay packages didn't look all that competitive. It's possible that there are grad students who have family support.