Second, I find that the race and gender composition of students in a master’s degree programs is correlated with the typical earnings of the program’s graduates. All else equal, programs with larger concentrations of female and black students have lower earnings, on average. /5

There are two main findings. First, roughly 30% of all master’s degree programs would “fail” this test, suggesting that they leave their graduates worse off (in terms of earnings) after attending. The share of programs that fail varies greatly across fields of study. /4

Under this test, a master’s degree program would “pass” if its typical graduates earn more than bachelor’s degree-holders in the same field of study living in the same state, after also netting out the annual cost of the master’s degree program. Here’s an example: /3