Bryan Caplan  

Finishing an Econ Ph.D.: The Basic Facts

Krugman on Greece... Race and progressivism...
If aspiring econ Ph.D.s are going to read one article about their prospects for academic success, Stock, Siegfried, and Finegan's "Completion Rates and Time-to-Degree in Economics PhD Programs" (American Economic Review, 2011) is probably the single best piece to read.  Their survey included 586 Ph.D. students from the 2002 classes of 27 programs, and tracked them for a full eight years.  The raw results:


[Larger-print version of the table here]

Key observations: Completion rates vary heavily by school rank.  In top-6 programs, 33% finish in five years, and 75% finish in 8.  In programs ranked 7-15, rates are almost as high: 32% in five years, 71% in eight.  Outside of the top-15, rates crash.  Much but not all of the completion gap appears in the first two years of the program.

Of course, if you're contemplating a Ph.D. in economics, you won't be satisfied with simple bivariate results.  What happens if you regress completion probabilities on a wide range of traits?  The results are extremely messy.  There are so many independent variables and sufficiently few observations that almost nothing has statistically significant effects on completion - including school ranking. 

At least in my view, however, Stock et al. controlled for so many program characteristics - including "program-level two-year attrition rare"! - that it's hard to interpret this result.  I would have rather seen regressions that controlled for program tier (or precise rank), student characteristics, and that's it. 

In the absence of such regressions, here's how I'd interpret the results: If you're good enough to get into a top program, but choose a lower program, your odds of completion probably remain high.  But if, like most Ph.D. students, you plan to attend the best program that accepts you, Table 1 provides good estimates of your prospects.  Caveat emptor.

COMMENTS (6 to date)
JLV writes:

Stock et al. gets you through Grad School. For post-grad school success, we need to consult Conley and Onder ( Here, we find that the best students at lower ranked programs are much more productive in terms of publications than the median student at top places.

Attending the best program that accepts you might not be the best strategy, unless you are genuinely capable of being the best student in the best place that accepts you.

E. Harding writes:

Something interested and vaguely on-topic from Gallup:

Jack PQ writes:

It would be good to know: what drives attrition? Having been through an econ PhD program (ranked 13-18 depending who you ask), I can say that between one-third and one-half flunked out because of the comprehensive exams. Stories I've heard suggest that flunking comps is common in programs ranked 13-25, but rare in top 10 (McCloskey says nobody flunked out of Chicago when she was DGS).

So, how much is forced attrition and how much is by choice? After years 1-2, some quit due to opportunity cost ($$ prospects), others because they find out they hate doing economic research, etc.

Lauren writes:

Jack PQ writes:

McCloskey says nobody flunked out of Chicago when she was DGS).
I think you were writing about what Deirdre McCloskey wrote in her Fall 2000 Eastern Economic Journal article, online at about being Director of Graduate Studies at the U. of Chicago in the 1970s.

To be clear, Deirdre wrote that almost nobody officially flunked out. Of course, if you can't get anybody to be your thesis advisor, you might as well have flunked out. Unless you passed the Core exam (which at least at the time was different from the later, subject-specific Prelims), you pretty clearly were going to have a well-nigh impossible time finding anybody to be your thesis advisor. And you certainly weren't going to get any funding for your second year!

You were allowed to re-take the Core exam, so, sure, it wasn't like the 40 to 50% who failed the Core officially flunked out. But most who failed the first time typically got the message either because there were obviously too few professors to go around when it came to supervising students who hadn't passed, or because non-passing students simply got no funding. No funding was a pretty clear incentive to leave, unless you truly believed that you could pass the next time around or that you had the encouragement and trust of some specific professor. Which professor could make that clear by directly funding you. (Extremely rare, but it did happen.) Most students at the time had already themselves paid for their first year--tuition, living expenses, everything. Investing in your human capital with your financial capital is great, but at some point the costs can start to outweigh the benefits. Should I double down and pay for a whole second year on the off chance that I won't pass the Core a second time, or get the thesis advisor of my dreams? Spend a whole year from something I could otherwise do--attend a different school or go into the workforce? Grad students, like anyone else in an investment/employment situation, do best to attend to the odds and signals they are getting, and to assess their performance relative to those odds and signals. Which is quite consistent with Bryan's and Deirdre's arguments!

Jack PQ writes:

Thank you for correcting my statement, which was indeed imprecise and particularly lacking in context.

What I meant to say is, because expectations do not decrease that much as one goes down PhD program ranks, it is no surprise that a greater number of students are forced to/encouraged to quit. Nearly all Chicago or MIT PhD students can handle Stokey-Lucas-Prescott, I'm sure, but the same isn't true in a program ranked 25th. The reason I say this is because math proficiency (courses taken and GRE score) weighs heavily in admissions decisions.

Lauren writes:

Hi, Jack PQ.

You've made an astute and subtle point when you said this:

[B]ecause expectations do not decrease that much as one goes down PhD program ranks, it is no surprise that a greater number of students are forced to/encouraged to quit.
The only thing I want to add is that as you go down the Ph.D. program ranks, schools typically offer more and more funding to prospective grad students. They also offer more and more explicit or implicit guarantees to reduce the risk taken on by a grad student--such as the school's paying all expenses the first year, guaranteeing that funding that won't disappear after that first year even if the student is having trouble, guaranteeing that there are plenty of thesis advisors to go around so no reasonable student will be left high and dry, or emphasizing that there is a culture of taking into account individual abilities and working to develop those abilities. And also emphasizing that there are job placement opportunities and a structure for helping you get a job at the other end, even if that other end does not involve your finishing with a thesis.

The Stock, Siegfried, and Finegan table Bryan reproduced could perhaps use an additional few lines such as "Percentage of tuition/living expenses paid by the school"--by year (1st year, 2nd year, 3rd year, etc.). How can anyone talk sensibly about how long it takes to complete a degree without talking about the funding involved?

It's all about the funding.

When you get to the grad school level, signals are sent by how much funding you get. It's not all or only about what school you get into or which school you commit to go to. A grad student has to balance the funding offers with what is potentially gained in human capital, and also with alternatives and risk-taking on many fronts.

Hey, not much different between non-acadamic jobs and academia, eh?

Comments for this entry have been closed
Return to top