Arb Research

Showing the messiness of income effects of school quality

Your prior on the causal effect of school quality

Your trust in top quasi-experimental results

Don't take these results too literally.

The point is not to make a strong claim about the effect size (or functional form): instead we want to show you how much it depends, how unpersuasive the existing evidence is.

Caveats

Caveats

The above attempts to control for ability bias (i.e. the spurious selection effect of colleges picking people with large earning potential); we ignored studies with no causal identification.

It's just income, because that's what almost all economists study. There's good reason to think that non-income effects are larger;
no particular reason to think it's linear;
the evidence is heavily tilted towards the US;
the effect probably differs a lot by subgroup (e.g. working-class kids).
We're using average student SAT as a proxy for college quality. But no proxy is very convincing.
Owing to limited data, our question is about the attendance effect, not the graduation effect.
This is the effect after a 10+ year followup; the initial effect is larger.
We look within college-goers; the graduate / nongraduate effect is different.
These are point estimates, but our uncertainty is large; see main table in the report.

Notes

% effect on income (vs mean school) = increase in income for attending a school with average student SAT of x, over one with SAT 1100.
Average freshman SAT = mean SAT of students who start freshman year at that school.

Mean intake SAT of all schools is 1140. MIT mean is 1545. You can get average intake SATs from the College Scoreboard.
(Recall that the hard ceiling on the +400 SAT effect is +110% - and that this could only be realised if admission selection and signalling did literally nothing.)