It is an empirical question, but the responses so far have been deductive, and deductive from rather general premises about present and past economies. Moreover, they all fit a very traditional pattern: Things were different back in "my day."
The response may be correct for all I know, but I haven't seen any concrete empirical answer yet. How 'marginal' an existence are we talking about? One undergraduate professor at Western Michigan had gotten his Ph.D. during the Depression at Chicago. He got a job as a dorm director or something like that. I don't remember now exactly how he got his meals. His point, though, was that being a grad student in the Depression was in some ways better than any alternative employment he might have gotten.
The population we're talking about, past and present, were/are also mostly unmarried or at least no children. Floaters as it were.
Carrol