Suppose that Rochester is the 20th-ranked economics department in the United States. Further assume that the admissions process does a pretty good job of sorting the "best" students - the most able, most driven, whatever - into the "best" schools.
Now, making the reasonable assumption that talent is distributed evenly across academic disciplines, I count the following subject areas: biology, bioscience, chemistry, computer science, economics/business, engineering, health science, history/geography, geology/environmental science, language/linguistics, law/journalism, math, music/arts, philosophy/religion, physics, psychology/political science, sociology/anthropology/gender, ethnic and cultural studies, writing/literature.
Now, obviously you can quibble with my attempted division of all graduate areas of endeavour into equal-size groups, but let's say there are 17 other areas of roughly equal size to economics and business doctoral students. Finally, in our long list of huge assumptions, let's say the average yearly entering doctoral class in each of these areas is 25.
Then 19x17x25 = 8075. So this year's class at Rochester are on average about the 8000th best commencing graduate students in the US this year.
Note that we don't need to be too concerned about variance, unless it's very heteroskedastic, because underachievers/those improperly sorted into top programs will be cancelled out by overachievers/those improperly sorted into lower-ranked programs.
Now, I'm tempted to argue that economics students have on average better GRE scores than most other subject areas, but a lot of the listed subjects don't require the GRE...we'll leave it for now.