One thing I used to teach back when I taught, and which I would teach a bit differently now that I am old and wise, is how people only need to be a little bit racist (for example) for there to be big economic impacts of that racism. Weirdly I would say that back then the general point being conveyed was "sorry about your lot, but people aren't really all that racist!!!" instead of the more important point that it only takes a tiny bit of racism (or misogyny or...) for there to be big outcome differentials.
The rough economist logic is if the marginal benefits of trying harder are lower, you won't try as hard. Discontinuities in credentialing make this worse. Why get a Ph.D if you have a 20% harder time getting a job?
The important point is not "wow these shitty outcome differentials don't mean that people are really racist." The important point is "even a tiny bit of racism matters a lot."
