Why hire a postdoc ?
This is a question to which Principal Investigators (PIs) must be able to provide a convincing answer for themselves, as they assess the wisdom of applying for research funds and/or setting aside funds already available, for that purpose.
In an old post I offered my reasons for regarding a postdoctoral appointment as generally a good opportunity for a junior scientist, for professional and personal growth. But what’s in it for the PI, really ? There are, in my view, good and bad reasons for hiring a postdoc.
Let us begin by re-hashing the most common bad ones:
1) A postdoc is essentially an advanced graduate student (I have heard this statement not only from colleagues, but even from graduate students), who can be assigned more challenging tasks than a graduate student because (s)he is more mature and has already “proven” him/herself. Equivalently, a postdoc can join a group and be a productive contributor right away, without the initial learning curve of a graduate student.
2) A postdoc is a junior scientist who is eager to obtain some important result in a relatively short time, in order to boost his/her own chances of landing a faculty appointment. Consequently, (s)he can be expected to work hard and generate quickly publications that will enhance the visibility of the PI (the assumption being that the PI, by paying the postdoc’s salary, is entitled to co-author each and every article arising from the postdoc’s work).
3) A postdoc is experienced enough to contribute to the supervision of students, as well as to the writing of articles, research proposals, research reports, organization of conferences, and maybe even to teach, thereby freeing some of the PI’s limited time.
What is wrong with all of the above ? In practice, everything described above can and does happen, often even if the PI had not really planned for it in advance. However, to bring in a postdoctoral scientist mainly with the above objectives in mind is misguided, and will likely create undue expectations on the part of the PI, disappointment and missed opportunities for the postdoc, and possibly incomprehension and tension between the two .
To begin with, a postdoc is most emphatically not an “advanced graduate student”. Postdoctoral scientists need to establish for themselves a record of original, largely independent scientific work. The research work that one carries out as a graduate students is supervised (at least in theory, and therefore in the judgment of an external reviewer who does not know the person). Thus, the young person’s ability to perform independent research, lead a group, generate original and interesting problems on which (under)graduate students can be engaged, can only be cogently assessed by providing him/her with the opportunity to work under limited supervision on the part of the PI (much less of the doctoral major professor).
A PI intent on hiring a postdoc with the aim of assigning to the person some long standing “pet” problem, without any input on the postdocs’s part, or of inserting him/her into an existing, well-oiled “assembly line”, with little room for creativity and for original contributions, will do a disservice to the postdoc, as such a plan is not conducive to the goal stated above. That kind of program represents almost a continuation of graduate school, and fails to appreciate the maturity and experience of the junior scientist.
Ideally, instead, the postdoc will be given a lot of latitude in the choice of problem, as well as on the investigative strategy to adopt (obviously, compatibly with what the PI can put at the disposal of the young colleague, by way of expertise, computational resources, laboratory equipment and manpower etc.).
I do not know how many PIs really hire postdocs with the aim of beefing up their own publication record, essentially by piggy-backing on the work of the postdocs — it sounds like an urban legend to me. Anyone thinking of doing that, imagining that peers will not be able to identify the person whose crucial contribution is behind the work, is seriously, how shall I put it, “misunderestimating” the collective judgment of the community. And most PIs with enough money to hire a postdoc usually do not have much trouble generating their own publishable work.
Many PIs insist to co-author the work of their postdocs for two reasons:
1) For the purpose of showing progress on a grant renewal application. However, it is actually not necessary that the PI be an author, in order for a published piece of work to be counted as part of the research effort associated to a specific grant (as long as grant support is acknowledged at the end of the manuscript) 
2) A PI not having co-authored even a single paper with his/her postdoc may not be deemed a credible reference. I personally tend to regard this as a valid argument, which is why I think it is a good idea if my postdocs publish at least one paper with me as a co-author. Otherwise, the fact that their salaries comes from my grant does not, in my opinion, entitle me to co-author their articles. It should be done only if I have contributed significantly, and in any case, if they publish it, it must be mainly their work, not mine.
Finally, taking on major student supervising responsibilities, engaging in extensive proposal writing, teaching etc., are all activities that might, occasionally have a place in the course of one’s postdoctoral appointment, but only to the extent that they can be seen as directly benefiting the postdoc. A PI ought not dump on a postdoc undesired teaching or administrative duties (conveniently labeled as “valuable experience”), ask a postdoc to squeeze some work out of an unmotivated graduate student, and generally otherwise divert his/her focus off research in order to do work that really belongs with the PI .
OK but then… what’s in it for me ?
So, if a PI cannot demand that a postdoc slave away with the other postdocs and graduate students in order to help the PI stay ahead of the competition, write the PI’s proposals, supervise the PI’s students and grade the PI’s papers… well, then, what is a postdoc good for ? I mean, am I hiring these persons for my sake or theirs ?
Well… theirs. That’s right.
The most important reason why a PI should try to get funding and hire a postdoc, is because (s)he is supposed to. The reason why we can practice our favorite profession is precisely because at some point some PI gave us the opportunity of showing that we could do science on our own (that, and a huge amount of luck, of course). The least that we can do, in order to show some gratitude and keep the enterprise moving, is to provide the same opportunities to other with the desire to pursue a career in the sciences.
But there is more than that. There are, in fact, tangible benefits to having postdocs in a research group, although they ought not be measured in terms of “that many extra papers”, or “help with supervision of graduate students” etc.
Postdoctoral associates contribute to the intellectual activity of the group, by bringing in not only their technical ability, but also their own original ideas and their knowledge of their field (of whose recent literature they are on top).
Because of their younger age and enthusiasm, postdocs can very often play an important role in the education of graduate and undergraduate students, not so much by way of teaching them “lab tricks” and helping them in their research (although that obviously happens), but more importantly by sharing their experiences and observations with them, thereby proving an invaluable source of information and insight on which students can base their own career decisions .
From the standpoint of a PI, I am convinced that a solid record of advising postdoctoral associates who then go on to successful careers (either in academia or industry, either in research or not) does with time enhance one’s prestige and standing in the community .
Postdocs also represent a tremendous resource for the scientific community as a whole. Because they spend time at different places and travel extensively (to go conferences and to deliver seminars at universities), they end up functioning as open channels of communication, allowing for exchange of information between different PIs, departments, laboratories.
 Standard disclaimer: things vary across disciplines, and clearly there may be individual situations in which some of the recommendations that I make may be inapplicable and/or impractical. Still, whenever possible I think some consideration should be given to the criteria listed above, when envisioning the future work of a postdoctoral fellow.
 Any program director will give due credit to a PI capable of running a functioning, intellectually stimulating group, which generates interesting discoveries by virtue of the vigorous exchange of ideas and synergistic interaction among its members.
 As I have expressed before, I am very skeptical about the wisdom of having a postdoc invest in activities such as teaching a significant fraction of a short appointment, whose purpose should be first and foremost to build research credentials. I know that there are those who feel that this may be beneficial, in case the person is aiming at a teaching position (e.g., at a four year college) with little emphasis on research. My experience is that one good additional publication carries someone further than a semester of teaching. But, mileage may vary, of course.
Of course, a postdoctoral fellow should be expected to produce a good first draft of his/her own papers to be submitted for publication.
 A postdoc has an understanding of the dynamic of the job market that a PI, especially one who has not been looking for a job for a while, rarely can match.
 In more concrete terms, supervision of postdoctoral associates is also normally valued by funding agencies in the broader context of training of highly qualified personnel (see, for instance, here).