A Note on Why Quarter of Birth is Not a Valid Instrument for Educational Attainment

摘要

In their justification for using entrance cutoff dates and compulsory education laws as a natural experiment, the authors of Angrist and Krueger (1991) rightly give much attention to the effectiveness of compulsory attendance laws. However, the authors do not give proper attention to the decisions made by parents. If redshirting is commonplace and nonrandom, as it is in the ECLS-K data set, then the identifying assumption of monotonicity does not hold, and their identification scheme does not work. This problem is distinct from those discussed in Bound and Jaeger (2000).