Econ303300 Fall 2024
Professor Loretta Fung
Assignment 3: Multiple Regression Analysis and Dummy Variables
Due Date: December 2 (Monday) at 11:59pm
Note: Please submit the assignment before 11:59pm on December 2 (Monday).
Part 1: Exercises
1. [10 points] When using n = 100 observations to estimate the model:
y = β0 + β1 x1 + β2 x2 + u
you obtained SSE = 22, 250.5 and SER = σ
b = 20.50. Calculate:
(a) SSR.
(b) R2 and adjusted R2 .
2. [15 points] Consider a model in which the return to education depends upon the amount
of work experience (and vice versa).
log(wage) = β0 + β1 educ + β2 exper + β3 educ · exper + u
(a) Show that the return to another year of education depends on experience.
(b) We now estimate the model using data in WAGE2. The results are summarized
below.
log(wage)
d = 5.9495 + 0.0440educ − .0215exper + 0.0032educ · exper
(5.949) (0.0174) (0.0199) (0.0015)
2
R = 0.135, n = 935
(c) Based on the regression results, whether the returns on education depends on ex-
perience? (Hint: you can test whether β3 is significantly different from zero).
(d) For people with 5 years of work experience, what is the returns on education? Use
the exact interpretation to explain the percentage change of wage with an additional
year of education.
3. [10 points] An economist analyzes the cost of living (cost) in different regions in Taiwan.
Four regions in Taiwan are identified by using four dummy variables: N orth, Central,
South, and East. The following equation was estimated:
cost = β0 + β1 N orth + β2 Central + β3 South + u.
(a) (4 points) Based on the equation above, which region is the base group? What is
the difference in cost of living between East and North?
(b) (3 points) What is the predicted cost of living in North?
(c) (3 points) What is the predicted difference in cost of living between North and
South?
1
4. [20 points] Using a random sample from US CPS data in 1985, the following equation
was estimated:1
log(wage)
d = 0.561 + 0.089educ + 0.034exper − 0.0005exper2 − 0.118f emale + 0.133marr
(0.120) (0.008) (0.006) (0.0001) (0.066) (0.058)
−0.195f emale · marr
(0.082)
2
R = 0.306 n = 534.
The variable log(wage) is the logarithm of hourly wage; educ is years of education ;
exper is years of experience; f emale is a gender dummy variable; and marr is a dummy
variable indicating if the individual is married (marr = 1) or not (marr = 0).
(a) Is there strong evidence that exper2 should be included in the model? From this
equation, how many years of experience does it require to reach the highest hourly
wage.
(b) Holding exper and educ fixed, what is the estimated wage difference between single
male and married male? Express in percentage using the exact formula. How
statistical significant is this estimated difference?
(c) What is the estimated wage difference between single male and single female (in
percentage, precise formula)? Test the null hypothesis that there is no difference
between their wages, against the alternative that there is a difference.
(d) What is the estimated wage difference between single and married females (in per-
centage, precise formula)? What would you need to do to test whether the difference
is significant?
Part 2: Empirical Exercises
1. [25 points] Use data in WAGE1 for this exercise.
(a) [7 points] Use OLS to estimate the equation
log(wage) = β0 + β1 educ + β2 exper + β3 exper 2 + u
and report the results using the usual format.
(b) [4 points] Is exper 2 statistically significant at the 1% level?
(c) [6 points] Using the approximation
%∆wage
d = 100(βb2 + 2βb3 exper )∆exper,
find the approximate return to the fifth year of experience. What is the approximate
return to the twentieth year of experience?
(d) [8 points] At what value of exper does additional experience actually lower predicted
log(wage)? How many people have more experience in this sample?
1
Data source: Economics Web Institute, StatLib and Berndt, ER. The Practice of Econometrics. 1991.
NY: Addison-Wesley.
2
2. [20 points, 5 points each part] Use the data in GPA1 for this exercise. To determine the
effect of computer ownership on college grade point average (colGP A), we can estimate
the following equation:
colGP A = β0 + δ0 P C + β1 hsGP A + β3 ACT + u
where PC is a dummy variable that equals 1 if a student owns a personal computer and
0 otherwise; hsGP A is high school GPA and ACT is achievement test score.
(a) Report your regression results in the usual form. What is the estimated difference in
GPA between students who own a computer and those who do not own a computer?
Is the estimated difference statistically significant? Hint: the results have been
discussed in the textbook on pages 209 and 210.
(b) Add the variables mothcoll (= 1 if mother is a college graduate; 0 otherwise) and
f athcoll (= 1 if father is a college graduate; 0 otherwise). Report the results in the
usual form. What happens to the estimated effect of PC ownership? Is PC still
statistically significant??
(c) Test the joint significance of mothcoll and f athcoll in the equation from part (b).
Use F test.
(d) Add hsGP A2 to the model from part (a) and decide if the quadratic term is needed.
