Some anecdotal evidence I heard from the older generation is that post docs in the sciences and mathematics didn't really appear (in the United States) until the Eisenhower era. During the cold war there was a big push by the government to invest in basic research, so suddenly everybody got more money to hire more graduate students. Unfortunately, the faculty positions are not opening up as fast. So you have a problem with supply bigger than demand. The expectation was that (a) more professorships will be created and (b) due to the explosion of interest in basic research, partly driven by the invention of the atomic bomb, industry will start hiring more PhDs. So it was expected that the over-supply of PhDs is a temporary problem (so does not need "curing" at the source) and a simple stop-gap (the post doc position) is enough. 

How much of this is actually true, I am not certain.