Students and postdocs are always pissing and moaning about salaries. Heck, Residents think they should be paid more and work less. In my experience, they’re just too scared to speak up as much as the postdocs (which is two points for the postdocs, in my view).
There are lots of reasons for salary levels. One of the most commonly cited is that The Scientific Enterprise (as funded by NIH and promulgated by the greybeards and occasional bluehair that run Medical School Research) wants Cheap Labor. Another is that supply and demand, in this relatively free market situation (free market in the sense that your job and training are not assigned to you by some external force) will reach an equilibrium. More people want to be scientists/postdocs than positions exist, so the salary levels are reduced.
However, the sorting and selection* process has always occurred. Most of the people who finished my program, and continued in science went directly into jobs (it was an ecology/evolution department) not postdocs. They were not always top universities, but having a job was good. But it was hard going, and sometimes people delayed graduating cause they had no place to go. And some people dropped out, not by their own choice. In those days, such folks were frequently married women who said what the heck, I might as well have a family. I had two math phd student roommates. The year that each graduated they had 3 or 4 job offers, and at Ivy League, Stanford, Caltech, etc. One chose NYU because he wanted to be in New York. When I first learned about their options I was astounded. These guys were good, but not That Good. Its just that their selection point came at another time – there were always lots of asst prof/instructor jobs in math departments to teach calculus to all the bio majors – each dept hired 4-6 new faculty each year. Their cut point came after 2-3 years when maybe one of the 4-6 guys was kept around (the NYU guy was smart, he’s still a prof there).
One of the things that has changed, even in E&E fields, is that multiple postdocs have just pushed that selection point later in time. And postdocs are more saavy than first year PhD students. Is the solution to move selection earlier- take less grad students? Is the solution to ensure that PI’s only take as many trainees as they can find spots for? Certainly there is a lot of ego at a lot of levels driving this problem. Everybody always thinks that they will win the lottery if they just buy enough tickets (see: the A2 NIH proposal resubmission idiocy).
*Selection is a process that causes an end state, sorting. Sorting, for our purposes, is the differential survival of trainees. See Vrba and Gould, 1986