1 Question 1
2
1. 2. 3.
4.
A. Interpret the results.
B. Is the effect of IQ on earnings large?
C. Any concern on the model? What is the potential problem in interpreting the coefficient on IQ?
D. What is the definition of R2 and how is it different from an adjusted? Question 2
Sample Exam Question
April 25, 2021
You have obtained data of 3,000 randomly selected individuals and are interested in the relationship between weekly earnings and IQ. The regression, using heteroskedasticity-robust standard errors, yielded the following result:
Earnings = 239.16 + 5.20IQ, R2 = 0.10, SER = 421.44, (1) where earnings are measured in dollars and IQ is measured in units (e.g. 120) respectively.
Consider a firm-year panel regression takes the following form
ROAit = β0 + β1Investmentit + μit (2) where ROA is a measure of firm performance, Investment is firm’s investment (capital ex-
penditures).
1. Name two omitted variables that could cause a bias in estimating the β1. Clearly explain why you think such bias exists.
2. If one use the firm-fixed model takes the following form
ROAit = β0 + β1Investmentit + αi + μit (3)
What does this model do for you in terms of getting consistent estimates of β1 compared to model (??)? Please give specific example.
1
3.
3
• • • • •
Currently, the model uses heteroskedasticity-robust standard errors. Are there any potential problem with this treatment of standard errors? Give remedies and briefly explain why your treatment of standard errors is suitable for this data.
Contents that will NOT show up in the final exam
Week 7 – The DERIVIATIONS of consistency of IV estimates Week 7 – J-tests
Week 8 – The DERIVIATIONS of clustered standard errors Week 9 – Regression Discontinuity
Week 10 – Content after slides 12 (Trees, NN…)
2