It used to be that there was no such thing as a postdoc, and you would go into a faculty position directly after your PhD.
Obviously, you can claim sour grapes in my case, but - I finished up my PhD which was somewhat middling but I did accomplish some important things - and had two very productive postdocs in disparate fields. What I observed was, consistently, PhDs getting degrees that probably shouldn't have. About half of the PhD thesis defenses I went to at The Scripps Research Institute (consistenly ranked in top 5) were, in my opinion, undeserved, possibly including mine. The situation was somewhat worse at the University of Maryland where I did a postdoc. My second postdoc was at the Venter Insitute, where there were no PhDs; we had a grad student intern in my lab that my boss selected that couldn't get a simple molecular biology procedure to work. The undergrad with no lab experience that I trained did it 50 times in 1.5 months. The grad student got his PhD, and is now a postdoc at LBNL.
Opinion aside, fact of the matter is that the academic system is broken for a simple, mathematical reason. There is an 'exponential growth' mechanism where you have professors with labs with, say, three postdocs, and four grad students (note that the grad students stay for longer than the postdocs). You can't expect to 'copy/paste' this tree indefinitely, even with a modest attrition rate, because ultimately this growth is unsustainable.
You have to be willing to cull, and cull hard. Confounding factors include: an unwillingness to cull because it makes the program look bad; sometimes you have to grant the PhD because you're afraid the grad program will be sued (insert anecdotal story of a student who nabbed a nice severance including a trip to Albert Hofmann's 100th birthday party); PhDs being sold as 'the next step' for listless students who don't know what to do with their life next because all they know is schooling; PhDs having a 'social status' cachet; nationalistic measuring contests where countries try to project greatness as measured by how many PhDs they produce - aka race to the bottom.
The 'jobs outside of academia' for PhDs has basically come about as a relief valve for this exponential process and now that relief valve has likely been saturated.
That postdocs appear to be underpaid is a symptom of the market failure; there are too many of them around and so you can pay them less. Moreover, because so many crappy PhDs are produced, it becomes difficult to differentiate between them. This basically creates frustration for genuinely smart PhDs, and so they go into finance on wall street and cause stock market crashes by making the assumption that markets are normally distributed when they're actually levy distibuted.