For economists, that is. A superstar team of co-authors — Susan Athey, Larry Katz, Alan Krueger, Steve Levitt, and Jim Poterba — writes:
…students’ grades in required core courses are highly correlated across subjects. The Ph.D. admissions committee’s evaluation of a student predicts first-year grades and Ph.D. completion, but not job placement. First-year performance is a strong predictor of Ph.D. completion. Most importantly, we find that first-year Micro and Macro grades are statistically significant predictors of student job placement, even conditional on Ph.D. completion. Conditional on first-year grades, GRE scores, foreign citizenship, sex, and having a prior Masters degree do not predict job placement. Students who attended elite undergraduate universities and liberal arts colleges are more likely to be placed in top ranked academic jobs.
Here is the paper. This bit comes at the end:
Our results raise an interesting question: Why are some characteristics much stronger predictors of grades than of job placements? Foreign-trained and male students achieve substantially higher first-year grades, on average, but do not appear to be placed into much higher ranked jobs.
In my time, twenty years ago, foreign-trained students are more likely to have already seen the core material. It may also be that many elite, non-U.S. educational systems are better geared toward producing good grades than producing independent researchers.