Prof. Bryan Caplan

bcaplan@gmu.edu

http://www.gmu.edu/departments/economics/bcaplan

Spring, 1999

HW#3 (due April 12 in class)

Johnston/DiNardo:

Sample problems: (answers pre-provided!): 9.5, 9.6, 9.7, 9.10

Math problems: 9.8, 9.9 (a and c only), 9.11, 9.12, 9.13

Additional Computing problems:
Download the new workfile from the webpage. This file has three data sets on it.

Data for problem #1. 100 observations of Income, Education, Erredu1=(Education + Error1), and Erredu2=(Education + Error2) where Error1~N(0,1), and Error2~N(0,10).

Data for problem #2. 100 periods of price (p1) and quantity (q1) data for the model:

(S) Q_s=a+bP+u1

(D) Q_d=c-dP+u2 where a=0, b=3, c=50, and d=1.

Data for problem #3. 100 periods of price (p2), quantity (q2), and weather (w) data for the model:

(S) Q_s=a+bP+eW+u1

(D) Q_d=c-dP+u2 where a=-10, b=10, c=200, d=4, and e=5.

1. a. Regress Income on a constant and Education.

b. Assume you observe Education with error, so you only observe Erredu1. Regress Income on Erredu1.

c. Repeat (b), but now assume you only observe the noisier data Erredu2. Compare your results for (a), (b), and (c). What has happened to your coefficients?

2. Regress q1 on a constant and p1. What is the resulting coefficient? What econometric mistake have you made? Explain.

3. a. Regress q2 on a constant and p2. Does this give you a sensible estimate of the demand equation?

b. Use the presence of the weather variable to apply IV/2SLS to the demand equation. Comment on your results.