I. Discuss why the IVs in the studies below might not be valid:

  1. A study looks at the effect of attending catholic schools on student performance. The authors have cross-sectional individual-level data with students from different neighborhoods. They use the percent of the local community’s catholic population as the instrument for whether the student attends catholic schools.
  2. A paper looks at the effect of the number of siblings on children’s education level. The authors use data from rural China, where there is a strong preference for sons. They observe that families that previously had more daughters tend to continue giving birth to more children. Therefore, they use the ratio of daughters as the instrument for family size in the first-stage regression and regress the average education level of all siblings on predicted family size in the second-stage regression.
  3. Gary, Indiana, is a major steel production center in the world. A study tries to quantify the negative impact of China’s steel export to the US on the unemployment rate in Gary. There is a reverse causation problem because when Gary produces less, China can export more to the US. As a solution, the authors use China’s steel export to other countries as the instrument for China’s export to the US.

Sample Solution