Demand curves slope downward, even in the humanities

A series of new policies in the humanities and the social sciences at Harvard University are premised on the idea that professors need the ticking clock, too. For the last two years, the university has announced that for every five graduate students in years eight or higher of a Ph.D. program, the department would lose one admissions slot for a new doctoral student. The results were immediate: In numerous departments that had for years had large clusters of Ph.D. students taking eight or more years to finish, professors reached out to students and doctorates were completed.

No exceptions were made, and Harvard officials believe that their shift shows that there is no reason for a decade-long humanities Ph.D.

Here is more, I would say only that they did not go far enough.  You might think that this causes a lowering of standards, but I’ll predict that the people who leave sooner are no less smart (do some of you wish I had written "stupider"?) or accomplished than if they had stayed longer.

Elsewhere from, here is a study of economists and plagiarism.


What is the rate of return on spending 10 years in a PhD program?

Could it be that once you actually get the PhD, your standard of living goes down?

Well I don't know if people who leave earlier are stupider, but it is one of the cherished articles of faith in the Linux community that if Torvald had not been able to stay in grad school more or less indefinitely there would not have been a Linux.

Also on a personal note, my brother took about 2 decade in physics grad school to work his way to the elite fencing ranks and recently was ranked #1 in the country in his age class....

In my experience, the people who finish sooner are almost without exception better than the people that take longer. People who complete programs in 4 years walk out with a bunch of publications and good jobs. The people that take 10 years might get a job, but have crappy CVs in comparison. The reason people finish sooner is usually because they are better, more motivated, and have a stronger work ethic. These people would be even better, I suppose, if they took an extra six years, but it's better for them to spend that 6 years earning tenure rather than earning a Ph.D. (N.b.: I'm discussing philosophy Ph.D.s here.)

It is the same in my field (Epidemiology) -- people who take a very long time to earn a degree don't usually seem to benefit much from the extra time. Once you have the basic skill set, the costs of being a student really do seem to outweigh the benefits.

I am sure the advisers felt little incentive to part with a well trained and cheap lab technician who can teach undergrads while they themselves do their consulting gig.

A PhD is one of the most ill-defined projects a sane human being will
undertake. When is one done? The adviser must, by definition, have a
conflict of interest here. Esp. in the sciences. An eight year track
leaves the adviser with a student who has acquired the complete
skillset, takes no courses, and is most productive with minimum marginal
investment. Obviously they keep them in longer!

The cardinal principal of separation of duties has been grossly ignored:
The typical adviser performs many, mutually conflicting, duties: source
of funding, technical guidance, call the shots about "when a PhD is
finished". Far too autocratic and one-sided power distribution. Besides
the innate problem that the adviser and the student are typically
pandering to two disparate lobbies. The adviser to the academic
community of his peers and his funding agencies. The student to
potential employers. It might be quixotic to assume they are the same
but they aren't.

But, I'll still concede that the smart guys do get out earlier; but only coz' they learn to game the system earlier and better! lol


Thanks for couching this post in terms of the law of demand; in the "Concise
Encyclopedia of Economics," one can find a great quote from Stigler
indicating that should anyone find an exception to the law of demand he
would achieve "immortality" in the econ profession, and would be "rapidly
promoted." This jives with the matter-of-fact tone of your post, of
course. What is odd then, is that despite the demonstrated explanatory
power of this hypothesis to explain observable phenomena over the past
century, some of the top journals in economics have found it fashionable
in the past 15 years or so to feature publications that posit upward sloping
constructs called "demand curves" sans evidence.

indiana jim,

The truly upward-sloping demand curve has indeed finally been found.
Rice in parts of rural China, although the paper is still not published.
Ghost of a skeptical Stigler holding it back, I guess.

For the humanities, it is not just that the job prospects are not good,
but what jobs there are will likely be in much less interesting academic
environments and communities than the grad schools. So, as long as one
does not have children or other pressing financial needs, hanging on in
grad school can have a lot of year to year appeal. All the discussion of
exploitation of lab technicians and so forth is clearly relevant to the
lab sciences and not the humanities.

"Of course it's also the case that life as a Harvard graduate student is pretty cozy, but I don't think that's the primary reason why people have been staying so long here. It's a localized problem that affects only the fields with terrible job markets."

I can see why the job market has a lot do with this.

Being at Harvard, your at the center of academia with interesting students/professors. Also a nice city.

And then when you look for jobs in a tough field like the humanities, you have to come to terms with the fact that you might end up at a small school in North Dakota.

This tells me a PhD should not take long to get.

That is interesting Barkley, I'll see if I can track it down and give it a look-see; thanks for the heads up.


I've started reading the paper by Jensen & Miller that you recommended.

Thanks again,



You're welcome, :-).

Is this exchange why Tyler is telling Robin Hanson about our
inability to disprove that demand curves never slope downwards,
or is he being more generally methodological, the latter more
appropriate for dealing with Robin?


I have read the paper and I think it is very well done. Its authors argue that one reason that their findings are important is that they make the case for the utility maximization model generally; apparently the argument had been made that because Giffen behavior had not been observed the model was invalidated.

The authors, I think, do a good job of explaining that Giffen behavior can be expected only under a fairly restrictive set of circumstances (people have to be very "poor" but "not too poor", but that there might be as many as 90 million people in China in such circumstances.

Although the authors do not say this, an implication of their paper is that if per capital wealth for all the people of the world should ever rise sufficiently about "subsistence", then Giffen behavior would cease to exist with regard to the consumption of staples.

Obviously, Giffen "goods" are nothing like the so called "Veblen goods" ("fashion goods", "status goods")that have more frequently of late been the subject of up-sloping demand speculations in some top general interest economics journals, again, sans evidence.


A colleague, reacting to reading the article you recommended to me (that I sent to him), provided the following interesting example:

A friend of his was saving for a vacation. His reaction to a rise in the price of electricity, he said, was to go out less often for meals and entertainment (so that he could continue to put away the funds for his vacation). But the result of this was that he used MORE units of electricity as a result (using his electric stove and microwave more often, watching more TV, keeping lights on for more hours, etc). In the face of an increase in the price of electricity this person's reported reaction was to use more units of it. Sounds like "Giffen behavior" to me.


The idea of "Giffen behavior" (as opposed to Giffen good) is a nice contribution of the paper you recommended to me

Comments for this post are closed