statistics assignment it is

Econ 300 Fall 2016 Problem set 6

Due Thursday 12/1

Instructions: Please write/type your answers and hand in a hard copy before class starts. The total points for this problem set is 50. There is no STATA application in this problem set.

1. Explain using two graphs (first stage and reduced form) in the example of birth weight and test score why RD renders the causal impact of hospital intervention on test score.

2. a. In the example of birth weight and test score, assume there is no manipulation, draw a graph illustrating the histogram test.

b. Suppose there is hospital manipulation-certain hospitals may refuse to take in pregnant women who suffer serious illnesses (assume these women tend to give birth to babies just under 1500g). Draw a histogram test illustrating the manipulation.

3. a. Draw a covariate test (take the covariate of mother’s education level) showing the RD is valid.

b. Draw a covariate test of mother’s education level, showing the case when higher educated mothers tend to give birth to babies just above 1500g).

4. Why is RD a local average treatment effect? List two reasons.

5. The following question is adapted from Mastering Metrics “5.1 A Mississippi Experiment”. You are welcome to read the book as a background reference. Research question: Should the Fed lend money to banks or restrict credit during financial crisis? A natural experiment occurred in Mississippi during the 1930s. In November 1930, southern states experienced a serious financial crisis. The Atlanta Fed, running the 6th District, lent to troubled banks in the crisis. By contrast, the St. Louis Fed, in charge of the 8th District, restricted credit. The border between the 6th and 8th Districts happens to run through the middle of Mississippi.

a. What are the outcome and treatment variables? b. Based on the above graph, do you see a treatment effect? Is the timing right? Explain. c. What is the treatment and control group if you adopt DID estimation strategy? d. What is the identification assumption? Do you think it is valid in this example? Explain. e. Do you think lending money to troubled banks during crisis would help save them? Why? f. Now focus on years 1929-1931. Define Ys,t where s is district and t is year (e.g. Y8,1929 is the outcome in district 8 in 1929). If you simply do a before-after comparison between Y6,1931 and Y6,1929. Do you think it renders the causal effect? Explain. g. How can you make the naïve comparison (Y6,1931-Y6,1929) causal? (hint: use a control group) h. Write down the DID estimate using Ys,t notation