Because schools in areas affected by these natural disasters were disproportionately low performing schools, many students affected were not only forced to switch schools, but also ended up switching...

Because schools in areas affected by these natural disasters were disproportionately low performing

schools, many students affected were not only forced to switch schools, but also ended up switching to a

higher quality school. Sacerdote finds initial declines in test scores (likely due to the shock of

displacement), but “by the third and fourth years after the disaster, evacuees displaced [by these natural

disasters] see a 0.18 standard deviation improvement in scores.” Sacerdote also finds that the

improvements observed were “concentrated among students initially in the lowest quintiles of the test

score distribution.”

1. Define the outcome (Y), the endogenous regressor (X), and the instrument (Z) in the context of

this study.
2. Is this instrument relevant? Explain why or why not.
3. Is this instrument exogenous? Explain why or why not.
4. For what group of individuals is this study externally valid? List at least two groups for which this

study is unlikely to yield externally valid estimates.