As an answer to question “Seminal papers that later were proven to contain errors”, @snoram mentioned Angrist and Krueger (1991), and there use of quarter of birth as an (exogenous) instrument for education.

@snoram illustrates using (Buckles and Hungerman, 2013) how quarter of birth is in fact correlated with important characteristics of the mother.

In a comment, @snoram write:

“For the record I think Angrist and Krueger's work is amazing, but today we know better than to use quarter of birth (in the us at least and for this purpose) as an instrument. Science advances.”

My question : which variables are know to be better than quarter of birth as instruments for, say, “years of schooling” (or more generally “education”)?


2 Answers 2


Check out Estimating the Return to Schooling: Progress on Some Persistent Econometric Problems by David Card (2001) ECA

I know distance from school has been used in the past as an IV..although it has been criticized.

  • $\begingroup$ I appreciate the reference +1. If someone wanted to write a self-contained answer that would be even better. $\endgroup$ Commented Jan 14, 2016 at 20:32

Whether or not the instrument is better depends on what outcome you are estimating.

In any case, some instruments that have been used:

  • quarter of birth
  • distance to schools
  • parental education
  • spousal education
  • number of siblings
  • availability of library card at young age
  • changes in compulsory schooling laws
  • regional differences in availability of pre-schools

Twin studies are also popular.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.