Here is a new study by Valerie Michelman, Joseph Price, and Seth D. Zimmerman:
This paper studies social success at elite universities: who achieves it, how much it matters for students’ careers, and whether policies that increase interaction between rich and poor students can integrate the social groups that define it. Our setting is Harvard University in the 1920s and 1930s, where students compete for membership in exclusive social organizations known as final clubs. We combine within-family and room-randomization research designs with new archival and Census records documenting students’ college lives and career outcomes. We find that students from prestigious private high schools perform better socially but worse academically than others. This is important because academic success does not predict earnings, but social success does: members of selective final clubs earn 32% more than other students, and are more likely to work in finance and to join country clubs as adults, both characteristic of the era’s elite. The social success premium persists after conditioning on high school, legacy status, and even family. Leveraging a scaled residential integration policy, we show that random assignment to high-status peers raises rates of final club membership, but that overall effects are driven entirely by large gains for private school students. Residential assignment matters for long-run outcomes: more than 25 years later, a 50-percentile shift in residential peer group status raises the rate at which private school students work in finance by 37.1% and their membership in adult social clubs by 23.0%. We conclude that the social success premium in the elite labor market is large, and that its distribution depends on social interactions, but that the inequitable distribution of access to high-status social groups resists even vigorous attempts to promote cross-group cohesion.
You can think of this as another attempt to explain the relatively high returns to education, without postulating that students learn so much, and without emphasizing signaling so much. Going to Harvard is in fact winning access to a very valuable set of networks (which in turn is signaling as well, to be clear).
For the pointer I thank Tyler Ransom.