1.
(a) Would you use differences in differences or first differences when using independent cross sections from two years? What if you were using panel data?
(b) Suppose you want to estimate the effect of different variables on annual earnings and that you have a panel data set on individuals collected on January 31, 2000 and January 31, 2002. If you include a year dummy for 2002 and use first differencing, can you also include gender in the original model? Explain.

2.
Suppose that in 2013 a school district with two primary schools was given a sharpincrease in funding. Call the high schools North and South. The district decided that the funding would go to South, mostly to hire more teachers to reduce class sizes. To evaluate the effectiveness of the increased spending, the district collected random samples of students in 2012 and 2014 at both primary schools. The measure of student performance was the Key Stage 2 test score given to 10-11 year old students in the district.
(a) Let score be the Key Stage 2 test score. Write down a model that allows you to determine whether the additional funding led to improved test scores; which coefficient will measure the effect of the new funding?
(b) Will the model from (a) be biased if South gets an influx of betterstudents because of the increased funding? Explain how?
(c) What additional variables should you control for?

3.
Suppose you want to test whether worse economic conditions in a country increase thechances of a civil conflict. You have a sample of African countries for which you have data on economic growth, captured by the variable g, as well as information on the onset of civil conflict, as given by con.
(a) What other factors would you control for in the equation? (You shouldbe able to reasonably collect data on these factors)
(b) Write an equation relating g to con and the other factors you listed in part (a).
(c) Suppose that government institutional quality is an unmeasured factorin the error term in part (b). Is this likely to be correlated with g? Explain.
(d) Discuss the assumptions needed for variation in rainfall for a countryduring a given time to be a valid IV for g.

1. (a) Would you use differences in differences or first differen...