In other words:

How do professors profit if a larger (rather than smaller) proportion of entering PhD students in their department complete their degree?

(At my department at least, it seems like the only benefit is "warm-glow altruism". A professor gains no more from having more students at her department complete their PhDs than I do by having fewer stray dogs in my town getting killed by traffic. A professor's career prospects does not seem in any way affected by whether 80% rather than just 50% of PhD students complete their degree. Not surprisingly, the completion rate, at least at my department, is closer to the latter.)

Note: I am not asking whether it is a good idea to give professors stronger incentives to increase PhD completion rate. Neither am I asking for suggestions as to how PhD completion rates can be increased. Rather, I am asking what the incentives at present are (granting of course that this varies from place to place).

  • 8
    In my country (Europe) the number of PHDs awarded per professor is considered during promotions, since it is indication not only on professors' capability of doing research but how he helps other people (his students) to become researchers themselves.
    – Alexandros
    Aug 23, 2014 at 9:39
  • 12
    Usually what matters is the number of completed PhDs (not the percentage), and this matters a lot. However, given the resources that you have available (funding for PhD students, time to supervise them, etc.), if you want to maximise the number of completed PhDs, it usually makes sense to try to maximise the percentage of completed PhDs. So there is plenty of incentive to make sure that your PhD students complete their studies on time. Aug 23, 2014 at 10:12
  • 2
    Are you asking what incentive I have to increase MY completion rate or my colleagues completion rate?
    – StrongBad
    Aug 24, 2014 at 16:59
  • A bit ambiguous about whether it's external incentives, or internal. Aug 24, 2014 at 19:54
  • 1
    Your question seems to emphasise "department-level" completion rates. This is a performance metric for a head of department or head of research. However, for an individual professor, the relevant performance metric is the completion rate for the professor's PhD students. Aug 25, 2014 at 1:19

6 Answers 6


In the United States, the National Research Council (http://www.nap.edu/rdp/) regularly ranks university departments. One of the criteria by which departments (and thus universities) are ranked is the "Average of Median Time to Degree" with the presumption that lower is better.

In my university, this results in the provost's office putting pressure on us to reduce our time to degree through a number of measures, some coercive and punitive. Most of the time, the coercion and punishment is directed towards the graduate students (i.e., after the sixth year they lose their rights to certain fellowships, they have to pay more for their gym and library memberships, etc.).

But the department as a whole gets some attention. Our chair has been regularly sending out lists of our orphaned and missing ABD candidates (some now in their 12th year) and asking us to find and terminate them (either through graduating them, or formally asking them to leave the program).

I've also been putting some pressure on my peers through the admissions system by suggesting (since they are my colleagues, I have no power) that people with many students on the books should refrain from taking on new students until they clear their backlog more a bit.

I should note that I'm in anthropology where: 1) students tend to go into the field and disappear for quite sometime; 2) doctoral students are mostly useless to faculty for slave labor as most of us work individually and not in labs (with the exception of some of our archaeological / biological colleagues).


If an advisee leaves without completing a Ph.D., then the advisor can end up feeling like the whole process was a waste of time, since the goal of strengthening the research area by training a future contributor was not achieved. That's not fair or reasonable: the education and experience may be useful for the advisee despite not culminating in a Ph.D., and the opportunity to try was itself valuable regardless of how it worked out. However, these feelings are relatively common, and the advisor's investment of time and effort becomes an incentive towards higher completion rates.

This personal investment begins only when the student chooses an advisor, so the corresponding incentive is not relevant before that stage. On the other hand, spaces in graduate school are a limited and therefore valuable resource, so there is still an incentive for faculty not to waste them, even at the coursework stage.

What makes this issue tricky is that nobody can agree on what the ideal completion rate should be. To a first approximation higher is better, but this isn't a universal principle, and a 100% completion rate would arguably mean you're doing something wrong. Over the course of a difficult five-year program aimed at a narrow career, some people will legitimately discover that there's something else they'd rather do instead, and others will be prevented from finishing by factors beyond their or the university's control. The only way to enforce a perfect completion rate would be to refuse to admit anyone who seemed like they might not finish and then bully all the students into finishing regardless of whether or how their circumstances had changed. For example, you would never take a risk by admitting someone who seemed like they deserved a chance, and you would never gracefully accept a student's change of plans. Ultimately, the fundamental difficulty is that we can't necessarily distinguish between reasonable attrition and inappropriate attrition caused by unsupportive practices, and this makes the whole issue contentious.


In France, labs are assessed by the AÉRES (mirror) (~= Department of Evaluation of Research and Higher Education).

The evaluation criteria (mirror) show that the AÉRES prefer labs where PhD students don't drop out too often.

Since the AÉRES evaluation has an important impact, e.g. to get funding, professors have an incentive to increase the PhD completion rate (and they also have an incentive not to turn PhDs into some ever-lasting position: labs typically to try to graduate students in 3-4 years).

In the US - at least in my university - I'm not aware of any incentive, except that new PhD students will avoid professors who have the reputation of taking a while to graduate their students.


I'm going to answer the opposite question. What disincentives do professors have to increase the PhD completion rate in their department? I'm looking at the opposite of what you asked because I feel it will give us insight into your question.

In some fields, which I can only talk directly about the one I am familiar with - Compute Science, having students complete their PhD in a timely matter isn't a focus unfortunatly. Professors who are already established may already have a little army of PhD grads out there citing their work and growing their PhD empire. Which means that current or recent crops of PhD students under that professor sometimes get stuck. Why? Unfortunately one thing that PhD students are is cheap(ish), reliable labor.

A PhD student is going to work on the tasks or research set before them by their PI. They are going to do their best to complete that research - not just because it benefits their career but because failure in that arena may lead to disfavor or even being removed from their program. A professor can be assured that PhD students working under them will focus on their research of choice and will invariably cite or reference previous work done by the professor. Incidentally the PI is almost always an author on any paper that student publishes.

That's a bit cynical, I'll admit. But even beyond the pure cynicism of authorship farms and indentured servitude is the basic truth that it takes time to train someone to do research work at a high level. If I have been researching creating artificial butterfly wings out of volatile polycarbonates(or whatever) then it's going to take at least a year, if not more, to get a student enough information, enough skill and enough confidence to wind them up and let them go on important research.

Right now some of the other Academic posters are rolling their eyes. This sounds both incredibly cynical and a little bit tin foil hat ish. But the thing is... these are real problems that have been identified in at least a few schools I know of. You get some old, cranky, tenured professor who has dozens of students out in the world and who has gotten used to his or her status as king of the research mountain and all the sudden students start taking longer and longer to graduate. All the sudden that professor forgets what it was like to be a PhD student, trying to make a difference with your research and trying to get the heck out and graduate and only remembers that it's annoying and expensive to train a new PhD student and, btw, this student's producing pretty well and... well it would be more convenient if this trained student would stick around for a while.

What is done about this? Some programs tie tenure positions with completion rates and timelines. This is problematic and, to be honest, I've heard people talk about it but I've never seen or heard of it in action.

Some programs have a soft cap on the length of time a student can remain as a PhD student. I have seen this in action and I'm conflicted about it. On one hand, PhD students need to either graduate or leave the program eventually. But different fields can take different amounts of time to get results. Additionally, sometimes life just happens and the student who is moving along their degree runs into a snag(changes PIs, has a health emergency, just needs more time). The hope is that these soft caps are flexible enough to deal with these situations. The ways I have seen these soft caps(for lack of a better word) implemented is by having a fairly strict timelines for course work completion, having a deadline for committee selection(generally very early in the process) and having regular meetings between advisors(academic style advisors, not research advisors) and students.

Although your answer states that you're not interested in what could or should be done in these situations I think it's valuable to say that: I think low PhD graduation rates are signs of an unhealthy department and unhealthy academic culture. They should be taken very seriously and the fix should never be the 'warm fuzzy glow' of graduating someone.


There's a bit of an inaccurate presumption in the question itself, or anyway ambiguous wording. That is, some assumption of causality...?!?

As a mathematician, I do not think of PhD students individually or collectively as assisting my research program so much as doing teaching work that I don't have to do. Even my own PhD students, while stimulating to talk to, do not "help" my research program. Perhaps this is contrary to mythology. Although more-experienced TAs (teaching assistants, in the U.S.) are generally better at their job due to that experience, that level of proficiency is attained within a year or two of beginning the role. That is, in my sort of situation, there is absolutely no motivation for faculty to delay completion of PhDs. Indeed, as @roboKaren comments, there is some pressure from higher-ups to improve "stats". (That is, in terms of TAs, we cycle them through at a certain rate, so there're always roughly the same number to do the TA work. Shortening or lengthening the cycle time would accomplish nothing, and we have no motivation to do so.)

Is the supposed issue about people leaving the program? This is a very complicated question, considering that it might be entirely reasonable to leave a program if/when one discovers it's not what one thought. Should we brow-beat people to stay? Bribe them? True, being nasty to try to "drive out weaklings" is incivil, but I don't address that possibility.

If the question is about the "drop out rate", then it is probably impossible to answer usefully. If it is about time-to-completion, some useful things can be said.

This simplifies the question to two very different ones: about my interest in colleagues' students finishing faster/slower, and about my interest in my students finishing faster/slower. Apart from general humanitarian/professional-ethical concerns, I have no interest in the arc of other peoples' students, as it is really not my business. It is true that my perception (based substantially on two stints as Director of Grad Studies in Math) is that some PhD students and their advisors devolve into an unproductive, uncommunicative relationship, with "blame" not clearly assignable, but which creates enormous difficulties in completion. E.g., if either party is naturally non-verbal (despite the seeming ok-ness of this in math, ... which is not the case, actually), there is a potentially fatal problem. If there is a misunderstanding (from either side) about the context, origins, degree, and sense of the alleged "novelty/progress" of a thesis, this can be a huge obstacle.

For my own students, I try to arrange a thesis project that will optimally use the talent, energy, background, etc., in the time fully-funded by the department as TA (not counting on RAs (research assistantships), fellowships, etc). My thinking is that one's clock really, really starts once one is "out the door", so taking a sixth year instead of just five may be well worth it. But if/when someone says they want to finish more quickly, a suitable project can be chosen, up to a point: usually, there is a misunderstanding, a misguided prior belief that, indeed, somehow, faculty or departments are conniving to delay degree completion. :) "If only they'd get out of my way so I could do my research...!?!" ... overlooking the point that all the "required" stuff is meant to assist that (nevermind the common stylizations and distortions and misunderstandings).

So: incentive for quicker completion than the funded period? None, on scientific and educational grounds. On bureaucratic grounds, there's an ever-hardening soft cap on funding, which is a good reason to get people out the door... even though this soft funding cap is meant in practice primarily to avoid funding "eternal grad students" who make a bit of progress but either can't finish a PhD-worthy project or who are unhire-able for some reason (e.g., inability to speak in English, in the U.S.).

So, while my experience as Dir Grad Studies in Math made me unhappily aware of others' foibles, I don't think there's much motivation in mathematics for faculty to delay PhD graduations, and only a little to accelerate it. But faculty aren't the obstacle, in most cases. It's just that the U.S. undergrad system under-prepares people for grad school, so there're two or so years of basic coursework necessary to be even marginally literate, and then becoming aware of the basics of any fragment of modern mathematics can (obviously... though people in CompSci seem to have a radically different epistemological mythology) take a year or two. Unsurprisingly, quite a few people discover that "math" is not what they thought it was, given that U.S. coursework typically is at best very-early 20th century stuff, giving no inkling of mid-to-late 20th century stuff, apart from the not-really-contemporary-research, somewhat delusional (inevitably fairly contrived) "REU" (research experiences for undergrads) episodes. Given the popularity of REUs in the U.S., I'll not comment further on the alleged virtues-or-not... apart from agreeing that, yes, it is good to spend some time hanging out with other same-age kids who really like math, and not being in the usual stodgy classroom situation...

So, if "completion rate" means completions-per-admission... well, that's a can-of-worms, all the more for programs (such as ours) which try to be imaginative in visualizing the success of some students who do not necessarily have the traditional trappings of "success", but seem to have sufficient interest and talent. Gambles! If we'd be taken to task for not gambling on sure things... ?!? And a similar analysis applies to length of time to completion: pressuring people or limiting them to finish more quickly could induce more "failures", rather than "coerce" faster success.

Again, I feel that the question itself is predicated on a misunderstanding of why it is not so easy to get a PhD... but maybe I'm just old-and-tired. :)

  • This depends very much on your discipline. In many experimental sciences, for example, experienced PhD students are a lot more useful to the lab then newbies.
    – ff524
    Aug 26, 2014 at 0:36
  • @ff524, indeed, I know, but/and the question's context was not clear to me. Even in math, there seems to be a traditional mythology among grad students in which, somehow, faculty "keeping down" grad students mysteriously benefits the faculty... Aug 26, 2014 at 0:39
  • 1
    @paul garrett Like you, I'm a mathematician, and I have always wondered if we're producing more Ph.D's than we need. On the other hand, I don't want to discourage, or remove the opportunity, for anyone who really wants to pursue mathematics deeply from doing so. Solution? I don't know. Aug 26, 2014 at 2:08
  • 1
    @paul garrett (to continue) I agree that US undergrad doesn't prepare students for graduate work. I work at a less competitive liberal arts school so I see the problem first-hand. But then again, US high schools are apparently not preparing their grads that well for college either. It's a vexing problem for which there is no easy solution. Aug 26, 2014 at 2:12

PhD is largely depend on the individual effort and skill of the graduate student, much more than the result of a student in lower educational levels. Therefore it is questionable how much credit should be given to the supervisor and the individual.

Indirect intensives of being a good mentor is numerous from tenure track level, as it is mentioned in the comments. A strong prof-grad students relationship is also a very strong basis of future research networks for both the students and the professors. I saw several examples how a successful former graduate student can raise the reputation of a prof, and help in new cooperation.

Direct incentives (head money? ) to the teacher to improve that rate would be counterproductive. PhD is not elementary school: it is not like pass or fail. Just because someone passed it, the career prospects doesn't get much better if it is not coupled with actually skills, good publications etc. So just pushing someone to pass when it is not deserved doesn't do any good to students neither. Even if the reason of failure is beyond the student, or the prof is unreasonably demanding, changing the lab and / or institutional level of protection of the student is much better solution i believe.

Imagine this scenario: top notch laboratories can (should) have really challenging research topics. Big part of the research is however carried out by graduate students. Do we offer direct intensives to the professor to lower the level and bet on safe projects, just to have higher pass rates for his graduate students? Easy trick, yet at the end the student ends up with poor publication record and the once-prestigious lab lower its quality, too. It is not good to anyone.

  • 3
    Do we offer direct intensives to the professor to lower the level and bet on safe projects — Heh. You mean aside from grants?
    – JeffE
    Aug 24, 2014 at 9:29
  • I don't think that the grant system would perfect, but this imperfection is not good reason to make more poorly designed incentives.
    – Greg
    Aug 24, 2014 at 19:37
  • 2
    Oh, yes, I totally agree that we shouldn't make things worse. But if you think academia doesn't offer numerous direct incentives for faculty to bet on "safe" projects, you're not paying attention.
    – JeffE
    Aug 24, 2014 at 20:31
  • If you want to discuss grants and other elements of science policy, you may want to open a separate question or something. These comments are rather limited spaces for off-topic conversations.
    – Greg
    Aug 25, 2014 at 1:53
  • "PhD is largely depend on the individual effort and skill of the graduate student" While I would agree that the WORK and OUTPUT of a PhD is the result of the individual effort and output of the student I think this misses the actuality of the post graduate education system. Professors has a real effect on when a student can propose and defend and many programs will not allow a student to move forward unless their professor gives the OK. This removes a large portion of the autonomy of the student when it comes to the time line on which they graduate.
    – Nahkki
    Aug 25, 2014 at 12:49

You must log in to answer this question.