1. You have been hired by the World Bank to study the effects of an incentive program for girls education in a state in India on gender differentials in school enrollment. Your supervisor suggests that you do a difference-in-difference test across school districts in the state: a few school districts have recently implemented programs of the following form: parents of girls who attend more than 80% of the days in a month receive some money at the end of the month.
   1. What data would you need to implement this difference-in-difference strategy? Draw the 2 x 2 box that demonstrates this strategy. What assumption are you making here?
   2. How might studying school enrollment for boys separately in the same difference-in- difference box be a falsification test? What problems do you foresee with this falsification test?
   3. Examining the data more closely, you notice that after the program, school enrollment for girls is 12 percentage points higher in districts that implemented the program relative to those that did not. You also notice that before the program, school enrollment for girls was 5 percentage points higher in districts that implemented this program relative to those that did not. What is the difference-in-difference estimate? Do you see any problems with this strategy given this new information?