My impression is that "in the olden days", it was normal that the PhD thesis was the first publishable research one produced. But as academia became more and more competitive, many advisers became aware of the fact that for their students to be considered "doing well", they should publish before the end of their PhD. Some people publish preliminary results of their PhD research, only to quote them later in their thesis.
I would say that at a broad range of universities, advisers are looking to help graduate students publish as soon as possible. (But most advisers also know that giving a publication "for free" doesn't help develop the research attitude of the student.)
Still, I've seen in various places the practice of letting a student carry out some easy calculation, which becomes part of a more advanced paper, which the student may not actually fully understand. But still the student is listed as co-author. This is supposed to prepare the student for "research" and it may be considered part of the "learning experience" to present this paper at seminars or conferences.
In this way some Master's students have a (usually joint) publication with their adviser (and possibly other students). In particular, this means that, by year 2.5 of their PhD they have at least 1 or 2 papers.
This practice seems to be common enough for interviewers to not even ask about joint publications of graduate students (maybe unless they are in top journals). (I've been interviewed for funding and my interviewer asked me "So, I saw on your CV that you have a singly-authored paper..." ignoring my B-grade journal joint publication.)
In that way "The System" knows about smart advisers. Now, advisers are trying to help students prepare singly-authored papers (or "first-authored" papers, depending on the field). The motivational barrier is now much higher, of course, since the adviser won't be listed as co-author. One pay-off for the adviser would be to increase his citation count by having the paper cite the adviser's papers.
(My personal impression is that "paper count" is a very poor measure of an application, but I also think it often still serves as a "rough first approximation" to the research ability of a student.)
To sum up, I think by now it is rather uncommon for a graduate student to have no publication by 2.5 years PhD. Many Postdoc positions in my field (Mathematics) are not filled with Postdocs, but rather PostPostPostdocs. Having no publications, or even just one publication resulting from your PhD work, I'm guessing it might prove rather difficult to find a job in academia.