Hello, dear friend, you can consult us at any time if you have any questions, add WeChat: daixieit

ECO 499 

Spring 2023

Quiz 3

Part I.

Part II. (40 points)

Note: Answer on the WORD file. Copy and paste computer commands and outputs.  

“Example---.R” files are from Heiss.

1. Panel data:  DID model

a. Use data in Example 13.2: effects of education and gender on wages

b. Estimate the model

i. lm(lwage ~ y85*(educ+female) +exper+ I((exper^2)/100) + union

c. Find the following from the estimation results.

i. The return to education in 1978

ii. The return to education in 1985

iii. The gender differential in 1978

iv. The gender differential in 1985

2. Panel data  

a. Use the data in Example 14.4: Has the return to education changed over time?

b. The model

plm(lwage~married+union+factor(year)*educ, data=wagepan.p, model="within")

i. “married” and “union” are included. However, we cannot include variables such as “place of birth” or “a city’s distance from river, or even “educ”. Why not? Explain.

ii. How do we know certain variables are constant over time?

iii. The time variation in the effects of education on wage is captured by the interaction term. Has the return to education increased?

iv. Do the F test for significance of all 7 interaction terms.

· Degrees of freedom in the unrestricted regression? Its RSS?

· Degrees of freedom in the restrictions (or the numerator degree of freedom)? Its RSS?

o You should run another regression for this.

· What is the F statistic?

· What is the P-value?

v. This is a two-way fixed effects model. Are the time-fixed effects significant individually?

vi. Test the joint significance of the fixed time effects. (Follow the same procedures as in (iv).)

3. IV and 2SLS

a. Use the data in Example 15.7 and 15.8: return to education

b. Estimate the model:

log(wage) ~ educ+exper | data=oursample

i. Using OLS

ii. Using IV method with “motheduc” as IV

iii. Using 2SLS manually with “motheduc”, “fatheduc” and “huseduc” as instrumental variables.

iv. Using 2SLS automatically with “motheduc”, “fatheduc” and “huseduc” as instrumental variables.

c. Test whether “educ” in model (b) is endogenous using all 3 exogenous variables.

d. For the identification of the model (c),

i. How many instrumental variables are required?

ii. Test the overidentifying restrictions using all three exogenous variables as instruments.

4. IV: Weak instruments?  

a. Use the data in Example 15.3: birth weight and smoking

b. Estimate the model: lm( log(bwght ~ packs, data  = bwght)

i. Using OLS

ii. Using IV with cigprice as instrument

iii. Using IV with lfaminc (log family income) as instrument

iv. Using IV with motheduc as instrument

c. Organize the regression results using stargazer.

d. Compare the models. Which instrument will you choose? Why? And how?

i. lm( packs ~ cigprice, data  = bwght)