(It used to be 4 and 5.)
A student in her 5th year who doesn’t have a stellar job market paper is always tempted to stay another year and try to produce something better. This is the ex post incentive of an individual student.
But ex ante the department as a whole would like to enforce a commitment for all students to go on the market in 5, even those whose job market paper at that stage leaves something to be desired. The basic reason is risk aversion. Every year they spend in grad school they produce another signal for the market. Good signals improve their prospects but bad signals make them worse. They would avoid the additional risk by committing to stay only 5 years rather than 6.
Now consider a student whose job market paper in year 5 leaves something to be desired. If she stays another year and produces a good paper, then although she is better off, she raises the bar for her colleagues and thereby strengthens their incentives to stay another year. A department policy that strongly incentivizes students to finish in 5 is needed to prevent the implied unraveling.
But that’s MIT. Then there’s everybody else. Students in other departments have to compete with MIT students for top jobs. At a department like Yale, only the best students will be able to compete for top jobs and this makes them risk loving not risk averse. Instead of wanting to minimize signals, the best Yale students want to produce enough signals in hopes that at least one of them is good enough to give them a shot at a top department job.
So one should expect funding, TAships, and face time with advisors to drop after 5 years at MIT but continue into the 6th year at Yale. (Note: I have no data on this.)