ECO 311

Practice Question set 1

Course Instructor: Bertha Nguluwe

PART A: THEORY

QUESTION 1

Given the following regression model with all the Classical Linear Regression Model (CLRM)
assumptions holding.

𝑌𝑖 = 𝛽0 + 𝛽1 𝑋𝑖 + 𝜇𝑖

(a) Find the estimators for 𝛽0 𝑎𝑛𝑑 𝛽1 using OLS

2 2
1 𝑋̅ ∑𝑋
(b) Show that 𝑣𝑎𝑟(𝛽̂0 ) = 𝜎 2 (𝑛 + ∑ 𝑥 2 ) = 𝑛 ∑ 𝑥𝑖 2 𝜎 2
𝑖 𝑖

𝜎2
(c) Show that 𝑣𝑎𝑟(𝛽̂1 ) = ∑ 𝑥 2
𝑖

(d) Show that the estimators, 𝛽̂0 and 𝛽̂1 are unbiased.
(e) Show that the estimators, 𝛽̂0 and 𝛽̂1 are consistent.

QUESTION 2

Comment on the significance of the following Classical Linear Regression Model (CLRM)
assumptions:

(a) Zero mean of the errors.

(b) Homoscedasticity.
(c) Zero correlation between the error term and the regressors.

QUESTION 3

You are given the following regression with all the assumptions of the Classical Normal Linear
Regression Model (CNLRM) holding. The model is estimated using OLS.

𝑌𝑖 = 𝛼0 + 𝛼1 𝑋𝑖 + 𝜀𝑖

(a) Show that if 𝐸(𝜀𝑖 ) ≠ 0 then 𝛼̂1 is biased

(b) Suppose that 𝛼̂1 = 0, find the OLS estimator for 𝛼0
QUESTION 4

You are given the following linear regression;

𝑒
𝑌𝑖 = 𝛽0 + 𝛽1 𝑋𝑖 + 𝜇𝑖 ; 𝜇 = 𝑋2

Where e is a random variable with 𝐸(𝑒) = 0 𝑎𝑛𝑑 𝑣𝑎𝑟(𝑒) = 𝜎𝑒2 . Assume that e is independent
of X.

(a) Show that 𝐸(𝑢|𝑋) = 0, so that the key zero conditional mean assumption is satisfied.
[Hint: if e is independent of X, then 𝐸(𝑒|𝑋) = 𝐸(𝑒)]
(b) Show that the homoskedasticity assumption is violated. [Hint: 𝑣𝑎𝑟 (𝑒|𝑋) = 𝑣𝑎𝑟(𝑒) if
e and X are independent.]

QUESTION 5

Suppose you are given the following regression which satisfies all the Gaussian assumptions.

𝑌𝑖 = 𝛽0 + 𝛽1 𝑋𝑖 + 𝜇𝑖

You estimate the parameters 𝛽0 𝑎𝑛𝑑 𝛽1 using two linear estimators. Using least squares, you
obtain 𝛽̂0 and 𝛽̂1 while the other technique you obtain 𝛽̃0 and 𝛽̃1.

(a) Assuming that the least squares weight is w and that of the other estimator is v, show
the conditions that v must satisfy for 𝛽̃0 and 𝛽̃1 to be unbiased. Note 𝒗𝒊 = 𝒘𝒊 + 𝒛𝒊
(b) Show that 𝑣𝑎𝑟(𝛽̂0 ) ≤ 𝑣𝑎𝑟(𝛽̃0) and 𝑣𝑎𝑟(𝛽̂1 ) ≤ 𝑣𝑎𝑟(𝛽̃1)

QUESTION 6

Suppose you are given the following regression which satisfies all the Gaussian assumptions.

𝑌𝑖 = 𝛼 + 𝛽𝑋𝑖 + 𝜀𝑖

(a) Show that 𝛼̂ = 𝛼 + (𝛽 − 𝛽̂ )𝑋̅ + 𝜀̅; and deduce that 𝐸(𝛼̂) = 𝛼

1 𝑋̅ 2
(b) Show that 𝑣𝑎𝑟(𝛼̂) = 𝜎 2 (𝑛 + ∑ 𝑥 2 )
𝑖

(c) Show that 𝛼̂ is a consistent estimator of 𝛼

QUESTION 7

Consider the following two models;

𝑌𝑖 = 𝛽0 + 𝛽1 𝑋𝑖 + 𝜇𝑖
 𝑌𝑖 = 𝛼0 + 𝛼1 (𝑋𝑖 − 𝑋̅) + 𝜇𝑖

(a) Are the OLS estimators for 𝛽0 𝑎𝑛𝑑 𝛼0 identical?

(b) Are the OLS estimators for 𝛽1 𝑎𝑛𝑑 𝛼1 identical?

QUESTION 8

You are given the following linear regressions which satisfy the Gaussian assumptions

𝑌𝑖 = 𝛽0 + 𝛽1 𝑋𝑖 + 𝜇𝑖 (1)

𝑌𝑖 = 𝛼0∗ + 𝛼1∗ 𝑋𝑖∗ + 𝜇𝑖 (2)

𝑌𝑖∗ = 𝜋0∗ + 𝜋1∗ 𝑋𝑖∗ + 𝜇𝑖 (3)

Where 𝑌𝑖∗ 𝑎𝑛𝑑 𝑋𝑖∗ are standardized variables.

(a) Using equations (1) and (2), derive the relationship between 𝛽̂1 and 𝛼̂1∗
(b) Show that 𝛼̂1∗ = 𝛽̂0 + 𝛽̂1 𝑋̅
(c) Using equation (1) and (3), derive the relationship between 𝛽̂1 and 𝜋̂1∗
𝛽 ̂
(d) Show that 𝜋̂1∗ = − (𝑆 1 ) 𝑋̅
𝑌

̅
𝒀𝒊 −𝒀 ̅
𝑿𝒊 −𝑿
Hint 𝒀𝒊∗ = and 𝑿∗𝒊 =
𝑺𝒀 𝑺𝑿

QUESTION 9

Suppose you are given the following regression which satisfies all the Gaussian assumptions.

𝑌𝑖 = 𝛽0 + 𝜇𝑖

(a) Show that 𝛽̂0 = 𝑌̅

𝜎2
(b) Show that 𝑣𝑎𝑟(𝛽̂0 ) = 𝑛

(c) Show that ∑ 𝜇𝑖2 = ∑ 𝑦𝑖2 = ∑(𝑌𝑖 − 𝑌̅)2

QUESTION 10

You are given the following sample linear regression model with all the assumptions of the
Classical Normal Linear Regression Model (CNLRM) holding.

𝑌𝑖 = 𝛽0 + 𝛽1 𝑋𝑖 + 𝜇𝑖
Let 𝛽̆1 be the OLS estimator for 𝛽1 obtained by assuming the intercept is zero (𝛽0 = 0) and let
𝛽̂1 be the OLS estimator of 𝛽1 obtained by assuming the intercept is nonzero (𝛽0 ≠ 0).

(a) Show that the estimator 𝛽̆1 is biased.

(b) Under what conditions would 𝛽̆1 be unbiased?
(c) Show that 𝑌̅ ≠ 𝑌̅̆
(d) Show that the residuals 𝜇̆𝑖 are correlated with the predicted 𝑌̆𝑖

PART B: APPLICATION

QUESTION 11

The following results have been obtained from a sample of 20 observations on the value of
output (Y ) produced by a firm and the corresponding labour input (X).

20 20 20

∑ 𝑋𝑖 = 2,786 ∑ 𝑌𝑖 = 546 ∑ 𝑋𝑖 𝑌𝑖 = 68, 719

𝑖=1 𝑖=1 𝑖=1

20 20

(∑ 𝑌𝑖 ) (𝑆𝑥2 ) = 3, 218, 473 (∑ 𝑋𝑖 ) (𝑆𝑦2 ) = 132, 585.7

𝑖=1 𝑖=1

Where 𝑆𝑥2 𝑎𝑛𝑑 𝑆𝑦2 are are the sample variances of X and Y respectively

(a) Estimate the sample production function of the firm by OLS and interpret your results.
(b) Find the standard errors of the coefficients of the production function.
(c) What is the part of the variation in output produced which is not explained by the
regression?

QUESTION 12
A researcher is interested in examining the relationship between corruption and income
inequality as measured by the gini coefficient. He collects data from 6 countries. His
empirical model is given as follows: 𝑌𝑖 = 𝛼0 + 𝛼1 𝑋𝑖 + 𝜇𝑖
where Y =is an index of corruption, higher values imply higher levels of corruption and X =
is a gini coefficient. The sample data is summarised as follows:
6 6 6

∑ 𝑌𝑖 = 11.5 ∑ 𝑋𝑖 = 3.13 ∑ 𝑋𝑖 𝑌𝑖 = 7.38

𝑖=1 𝑖=1 𝑖=1
 6 6

∑ 𝑌𝑖2 = 25.75 ∑ 𝑋𝑖2 = 2.27

𝑖=1 𝑖=1

(a) What is the a priori expected relationship between Y and X?

(b) Estimate the corruption model and interpret your results.
(c) What part of variation in corruption is not explained by inequality?
(d) Find the standard errors of the coefficients.
(e) What is the policy implication of the results from the corruption model?
(f) Does this simple regression necessarily capture a causal relationship between
corruption and inequality? Explain.

QUESTION 13

Given the regression model;

𝑌𝑖 = 𝛽0 + 𝛽1 𝑋𝑖 + 𝜇𝑖

And the following information

Y 2 4 5 1

X 3 5 7 1

(a) Estimate the regression model by OLS

(b) Estimate the error variance (𝜎 2 )
(c) Estimate the variances of the OLS estimators in part (a) above
(d) Verify that the following equalities are satisfied by the OLS estimators
(i) ∑ 𝜇̂ 𝑖 = 0
(ii) ∑ 𝑌̂𝑖 = ∑ 𝑌𝑖
(iii) ∑ 𝜇̂ 𝑖 𝑋̂𝑖 = 0
(iv) ∑ 𝜇̂ 𝑖 𝑌̂𝑖 = 0
(e) What proportion of the variation in Y is explained by the model?

QUESTION 14

Redo question (13) using the following empirical information

Y 4 6 2 5 3

X 1 0 1 0 1