ETF2100/5910 Introductory Econometrics

Assignment 1, Semester 2, 2020

IMPORTANT NOTES:

• Type your answers using Microsoft Word or write your answers clearly. You must submit a
PDF file to Moodle. Other file formats are not accepted.

• Notation used in the assignment needs to be typed or written correctly and properly. Marks
are also awarded for presentation.

• When doing calculation, keep at least 4 decimal in each step for precision. For final answer, 3
decimal point is sufficient, unless specified otherwise in the question.

• In this assignment, when you need to use t critical value, you can either find it using Eviews
or use the statistical table.

• This assignment is worth 20% of this unit’s total mark.

• ETF2100 students must answer Question 1B and Question 2.

• ETF5910 students must answer all question (i.e., Question 1A, Question 1B, and Question 2.)

• Total marks for ETF2100 and ETF5910 students are 60 and 70, respectively.

• Marks will be deducted for late submission on the following basis: 10% for each day late, up
to a maximum of 3 days. Assignments more than 3 days late will not be marked.

Question 1A: FOR ETF5910 ONLY (10 Marks)

(a) Show that (5 Marks)

N N N N
X X 1 X X
(xi − x̄)(yi − ȳ) = x i yi − xi yi (1)
i=1 i=1
N i=1 i=1

(b) Show that (5 Marks)

N N N
!2
X
2
X 1 X
(xi − x̄) = x2i − xi (2)
i=1 i=1
N i=1

Page 1 of 4
Question 1B: For Both ETF2100 and ETF5910 (34 Marks)
*Please begin this question on a new page
In 2000, data were collected for 200 CEO Salaries (measured in millions of dollars) and their
company’s return on equity, RoE (measured in percent) over the previous three years. Consider the
following linear regression model.

Si = β1 + β2 Ri + ei (3)
where Si is the CEO’s salary, Ri is his/her company ROE, and ei is an error term, each for a CEO i.
You have no prior beliefs regarding the precise value of β2 . You obtain the following sample statistics:

200
X 200
X 200
X
Ri = 3110.9, Si = 500.3, Ri2 = 63786.9,
i=1 i=1 i=1
200
X 200
X
Ri Si = 10524.8, (Si − S̄i )2 = 419.2
i=1 i=1

(a) Using the information above and the fact that for any sample of data (xi , yi ) with N observations
the following relations hold:

N N N N
X X 1 X X
(xi − x̄)(yi − ȳ) = x i yi − xi yi (4)
i=1 i=1
N i=1 i=1
N N N
!2
X X 1 X
(xi − x̄)2 = x2i − xi (5)
i=1 i=1
N i=1

Calculate the LS estimates b1 and b2 . Be sure to state the formulas using the variable names
in this question (not just a generic y and x) and show your calculation. (6 Marks)

(b) Interpret the estimated slope coefficient. Be careful about the unit in your interpretation. (3
Marks)

(c) Interpret the estimate of the intercept. Does it make sense? Comment. (3 Marks)

(d) Suppose the estimate of the variance of the error term, σ̂ 2 , is 48.78. Compute the estimated
c 2 | R) and the standard error of b2 , se(b2 ). Be sure to state the formulas using
variances, var(b
the variable names in this question and show your calculation. (5 Marks)

(e) What does the size of the standard error of b2 tell us about our estimation of the relationship
between salary and ROE? (3 Marks)

(f) How does each of the followings affect the size of the standard errors of the least squares
estimators. Give a brief intuition.

(i) Smaller variance of the error term (3 Marks)

Page 2 of 4
 PN
(ii) Smaller sum of squares of the values of R about their sample mean, i=1 (Ri − R̄)2 (3
Marks)

(g) Economic theory states that return on equity should have a positive impact on CEO salary.
Test the hypothesis that the return on equity has no impact on CEO salary against the afore-
mentioned economic theory at 5% level of significance. Be sure to write down your null and
alternative hypotheses and show all the steps used to conduct your test. (6 Marks)

(h) Predict the expected CEO salary when the company’s return on equity is 12% over the previous
three years. (2 Marks)

Question 2: For Both ETF2100 and ETF5910 (26 Marks)

• S: Years of schooling

• SM: Years of schooling of respondent’s mother

• SF: Years of schooling of respondent’s father

We are interested in the relationship between individual’s years of schooling and his/her mother’s
educational attainment.
*Please begin this question on a new page

(a) Estimate the following linear regression model by least squares. Report the result and include
your Eviews output in your answer. (3 Marks)

Si = β1 + β2 SMi + ei (6)

(b) Why do we include the error terms ei in our model? Give two specific examples of variables
that maybe captured by the error term. (5 Marks)

(c) Interpret the estimated slope coefficient. Is this consistent with what you would expect the
relationship to be? Explain briefly. (3 Marks)

(d) Predict years of schooling for an individual whose mother’s education is 13 years. (2 Marks)

(e) What level of mother’s schooling predicts her child’s schooling of exactly 13 years? (2 Marks)

(f) Test at the 1% level of significance the null hypothesis that “an extra year of mother’s education
is associated with an increase in her child’s education of half a year.” using a two-tailed test.
Be sure to write down your null and alternative hypotheses and show all the steps used to
conduct your test. (6 Marks)

Page 3 of 4
(g) Construct a 95% confidence interval for β2 . Interpret the confidence interval. (3 Marks)

(h) If instead you are interested in studying the relationship between individuals’ education attain-
ment and his/her father’s educational attainment, what regression model would you estimate?
Write down the true model. No need to estimate it using Eviews. (2 Marks)

Page 4 of 4