University of Tunis 2020-2021
Tunis Business School Fall session
Econometrics
Tutorial N°3
Exercise 1
Consider the following model:
yi 1 2 x2i 3 x3i ui ; i 1,..., n
Suppose that: X e, x2 , x3 the observations matrix where e is the vector whose components
are all equal to one.
Given the following results:
23 230 115 4045 460 115 276
X T X . 2312 1158 ; X T X
1 1
. 69 46 ; X T Y 2770
. 460
. 587 . . 69 1388
y y 10
2
i
Part 1:
Compute:
x x2 , x x3 , x x2 x3i x3 , y , x x2 yi y
2 2 2
n, x2 , x3 , y , 2i 3i 2i i 2i
and x 3i x3 yi y .
Part 2:
We suppose ' ( 1 , 2 , 3 ) the parameters vector.
1) Determine ˆ , the OLS estimates of .
1
2) Determine the estimator of u2 .
3) Determine the coefficient of determination R 2 .
4) Test the global significance of the model at 5% risk.
Exercise 2:
Consider the model
Y 1 2 x2 3 x3 u
and suppose that application of least squares to 20 observations on these variables yields the
ˆ ˆ denote the estimated covariance matrix) :
following results ( Var
1 0.96587 0.21812 0.019195 0.050301
ˆ ˆ
2 0.69914 ,Var 0.019195 0.048526 0.030223
1.7769 0.050301 0.031223 0.037120
3
ˆ u2 2.5193 R 2 0.9466
1) Find the total variation (TSS), unexplained variation (RSS), and explained variation (ESS)
for this model.
2) Carry out the suitable testing procedures to test the individual significance of 2 and 3 at
5% level.
3) Carry out the suitable testing procedures to test the overall significance of the regression at
5% level.
4) Test the hypothesis H0: 3 1 against the alternative H1: 3 < 1 .
5) Test the hypothesis H0: 2 2 3 .
Exercise 3:
Consider the following model that relates the percentage of a household’s budget spent on
soft drinks SDRI to total expenditure TOTEXP, age of the household head AGE, and the
number of children in the household NC.
SDRI 1 2 ln TOTEXP 3 NC 4 AGE u
This model was estimated using 1200 observations. An incomplete version of this output is
provided below.
2
coefficients
Estimate Std.Error t-value Pr > t
Intercept 1.4515 2.2019 ? 0.5099
Ln(TOTEXP) 2.7648 ? 5.7103 0.0000
NC ? 0.3695 −3.9376 0.0001
AGE −0.1503 0.0235 −6.4019 0.0000
Residual standard error: ?
Multiple R-squared: 0.0575 Adjusted R-squared: ?
Residual sum of squares: 46221.62
1) Find the missing values that appear in this table.
a) The t-statistic for ˆ1 .
b) The standard error for ̂ 2 .
c) The estimate ˆ3 .
d) R 2 .
e) ˆ u .
2) Interpret each of the estimates ̂ 2 , ˆ3 and ̂ 4 .
3) Compute a 95% interval estimate for 4 .What does this interval tell you?
4) Are each of the coefficient estimates significant at a 5% level? Why?
5) Test the hypothesis that the addition of an extra child decreases the mean budget share
of soft drinks by 2 percentage points against the alternative that the decrease is not
equal to 2 percentage points. Use a 5% significance level.
Exercise 4:
Using a data of Tunisian citizens, we consider the following regression to explain Gasoline
consumption using income and price.
yi 1 2 x2i 3 x3i ui ; i 1,..., n
yi =Individual Gasoline consumption in liter.
3
x1i =Individual income in TND.
x2i =Price of liter in TND
ui =Disturbance term verifying the classical assumptions
We suppose that X (e, x2 , x3 ) the observation matrix where e is the vector whose
components are all equal to one.
Given the following results, where standard errors are given in brackets.
yˆi 27.3 0.510 x2i 0.350 x3i
(2.6) (0.2) (0.1)
n=23
R 2 93.7%
1) Construct a 95 % confidence interval for 2 and 3 .
2) Use the 5% significance level to test that
a) Individual income is a significant variable,
b) Gasoline price is a significant variable.
3) Use the 5% significance level to test the global significance effect of regression. Carry
out the suitable testing procedures. What conclusion can be drawn from the results?
