I learned some econometrics today. Allowing for heterogeneity with respect to how different individuals respond to different instrumental variables affects how we think about the LATE estimator.
aeaweb.org/articles?id=10…
aeaweb.org/articles?id=10…
Who is at the margin? Matt and Dora are both thinking of attending college. If they attend college, they will enjoy a treatment effect. Each of them recognize that attending college requires paying tuition and commuting to the school. Matt is rich but he lives far from school
Dora isn't rich and she lives close to the college. An econometrician who randomly subsidizes tuition and randomly moves the college will induce which of them to go to college? Who is "at the margin" when there are multiple price margins and heterogeneous program participants?
This AER paper has some cool sorting diagrams that look like something Sherwin Rosen would teach us about daredevils (shapes of indifference curves in risk and $ space) selecting to work in risky jobs. What "Z" would induce a wimp to work in the risky job? Roy Models!
I should edit this. Matt is rich and thus doesn't really care about the tuition. Dora has a fast motorcycle and thus she doesn't really care about her distance from the School. Dora will respond more to the tuition Z and Matt will respond more to the commute time Z.
• • •
Missing some Tweet in this thread? You can try to
force a refresh
