Even if you're not taking out a loan in the monetary sense, you're taking out a loan from your long term earnings, one that has little chance of being repaid, to put your mental assets/skills to a non-remunerable task, and get a certification that you did so. The problem is that the option to continue doing this (tenure-track academic jobs) are limited and (naturally) highly contested.
However, you make a good point--there's an underlying and insidious opportunity cost that is often unknowingly sacrificed: that of atrophying skill sets. It's easy for a PhD to be a hugely insular experience, if you let it, and if you take the easy way out and don't stretch your engineering skills (speaking in terms of CS here, since that's what I know), you're in for a rude awakening if you determine that academia is not for you. If you're not careful, you'll get good at writing papers, but might actually get /worse/ at writing portable, readable, and maintainable code. And as brilliant as your papers may be, if you can't ship good code, you're going to have trouble in industry.
The good thing, again at least for CS students like me, is that the "fun and enriching" environment of academia means a lot of opportunity for starting companies, creating libraries/frameworks, working on side projects, and doing contract work, so there's no reason you have to atrophy. Which is something that, sadly, the visions of the tenure-track academic job are engineered to beat out of you.
That may be so. Are you familiar with the concept of a wage being equivalent to the marginal product of labor (or value of last hour worked)? :).
I'm with you. I'm working on my dissertation at the moment.
1) It's quite common to study for PhD while getting paid for it instead of paying for it - and in that case, it should be treted as a fun and enriching (though not very high paying) job for a couple of years instead of "delaying 4 years". Getting into a huge debt for it is a whole different thing, though.
2) It feels that this problem is partly USA-specific. Sure, other places may have noticed similar tendencies but not as sharply; and science is very, very global - I'm seeing a lot of great researchers moving between countries/universities every 5-6 years based on where the major research projects in their area are happening & funded.
In general, a solution would be to try to decouple research from teaching instead of mixing them all under PhD/professor positions - the future tendency seems to be with much less people needed in teaching (due to changes in society, student funding/loan systems, MOOC's, etc) and thus more PhD's shifting towards research.