Midterm test. Econometrics.

Duration: 60 minutes. Only printed and handwriting documents are allowed.

Full name: ........................................................................... Student ID: .....................................
MULTIPLE CHOICE QUESTIONS
Choose or write down your answer after each question (explanation is not required). Calculation should be correct to
at least two decimal places.
1. Consider the following sample data:
xi - mean x -2.5455 -2.5455 -1.5455 -0.5455 -0.5455 0.4545 0.4545 1.4545 2.4545

X 4 4 5 6 6 7 7 8 8 8 9
Y 4.3 3.9 5.3 5.8 5.2 7.4 6.1 8.1 7.2 7.4 8.4
-1.9818 -2.3818 -0.9818 -0.4818 -1.0818 1.1182 -0,1818 1.8182 0.9182 1.1182 2.1182
The estimated linear regression model obtained is: tổng x = 72, y = 69.1
mean: x = 6.5455, y= 6.2818

A. Ŷ = 0.6519 + 0.8601X C. Ŷ = 0.8495 − 0.673X

B. Ŷ = 0.8601 + 0.6519X D. Ŷ = 0.7549 − 0.7732X
P50 P50
2. Consider a model with i=1 (yi − y)2 = 1085, i=1 (yi − ŷ)2 = 76. What is the coefficient of determination?
R2 = (1085 − 76)/1085 = 0.93.
3. Considering the model to estimate the relationship between sales (1,000 billion VND) and advertising (in billions
of VND), the estimated results are as follows: Ŷ = 35 + 7X. Based on this estimate results, if advertising is 10
billion VND, the point forecast value for sales (in VND) is:
105, 000 billion VND.
4. Consider the regression model between test scores (Y) and faculty/student ratio (X): Ŷ = 557.8 + 36.42 ln X.
According to this equation, when the faculty/student ratio increases by 1%, the average student test score
increases

A. 0.3642 points B. 36.42 % C. 36.42 points D. 557.8 points

5. A regression model with 34 observations, has a sample variance σ̂ 2 = 12.6 and ESS = 378. Then the number of
independent variables is:

A. 3 B. 4 C. 5 D. 6

Hint: σ̂ 2 = ESS/(n − p − 1).

6. Consider the regression model of output (Q) against capital (K), labor (L) and total factor productivity (TFP)
as follows:
ln Q = β1 + β2 ln K + β3 ln L + β4 ln T F P + ε.
One comments that when input factors such as capital, labor, and total factor productivity all increase at the
same given rate t > 0, output will not change. To test the above statement, hypothesis H0 is:

A. β2 = β3 = β4 = 0 B. β2 + β3 + β4 = 0 C. β2 + β3 + β4 = 1 D. β2 = β3 = β4 = 1

7. In a laboratory experiment, data were collected on longevity (Y in months) based on daily protein intake (X1 )
and a dummy variable X2 (whether a longevity extender be added to the diet of rats or not) with
(
0 if extender is not added,
X2 =
1 if extender is added.

From the experimental results, the regression model is obtained as follows: Ŷ = 36 + 0.8X1 + 2.7X2 . From these
results, the average lifespan of a mouse supplemented with a longevity extender every day and provided with 2
units of protein is:
36 + 0.8 × 2 + 2.7 × 1 = 40.3
8. Suppose we have a regression model of salary estimated based on experience and gender as follows:
wage = 15 + 13 exper + 2 gender,
where gender is a dummy variable with gender = 1 for men and gender = 0 for women. If we estimate another
regression model, with the gender variable reset to gender = 1 for women and gender = 0 for men, the results of
estimating the new model will be:
 A. Not change compared to the old model C. wage = 17 + 13 exper + 2 gender
B. wage = 15 + 13 exper - 2 gender D. wage = 17 + 13 exper - 2 gender

Hint: wage = 15 + 13 exper + 2 genderold = 15 + 13 exper + 2(1 − gendernew ) = 17 + 13 exper + 2gendernew .

9. Consider the regression results as follows:

Income = 11 + 0.24 GPA − 0.15 Female + 0.14 Married − 0.01 Married × Female,

where Female is a dummy variable for gender, with: Female = 1 if the interviewee is female and Female = 0 if
the interviewee is male; Married is a dummy variable for marital status, with: Married = 1 if the interviewee
is married and Married = 0 otherwise. Consider married individuals with equal GPAs. The above estimation
results show that compared to women’s income, men’s income is higher by an amount:

A. 0.01 B. 0.02 C. 0.15 D. 0.16

Use the following data for questions 10 – 13: Let Y : Chicken meat consumption/person/year (kg); X2 : disposable
income/person/year (dollars); X3 : retail price of chicken; X4 : retail price of pork; X5 : retail price of beef. From
the data obtained, we have the following estimated results:
Source | SS df MS Number o f obs = 20
=============+================================== F( , ) = 89.928
Model | Prob > F =
Residual | 21.127 R=s q u a r e d =
=============+================================== Adj R=s q u a r e d =
Total | Root MSE =

==============================================================================
Y | Coefficient Std . e r r . t P>| t | [95% c o n f . i n t e r v a l ]
=============+================================================================
X2 | 0.372 0.083 A 0.0007
X3 | = 0.292 B = 4.55
X4 | 0.128 0.099 C
X5 | 0.091 D 0.904
_cons | 2.189 E 14.602 0.0000
==============================================================================
10. Calculate A, B, C, D, E:
A = 0.372/0.083 = 4.482.
B = −0.292/(−4.55) = 0.064.
C = 0.128/0.099 = 1.293.
D = 0.091/0.904 = 0.101.
E = 2.189/14.602 = 0.15.
11. The 95% confidence interval of β2 (the coefficient of X2 ) is
0.372 ± t0.025 (15) × 0.083 = (0.21, 0.54).
12. Is β3 (the coefficient of X3 ) statistically significant at the 5% significance level?
Yes, because |t| = 4.55 > t0.025 (15) = 2.06.
13. Is the model statistically significant at the 5% significance level?
Yes, because F = 89.928 > F0.05 (4, 15) = 3.056.
14. Consider the following least squares estimation model:

ln Ŷ = 1.2 + 57.1 ln X.

According to this equation, how much does Y change when X increases by 1%, on average?
Y increases by 57.1%.
 CONSTRUCTED RESPONSE QUESTIONS

Write your detailed answer to the following questions. Calculation should be correct to at least two decimal places.
15. Consider the multivariate regression model

(U ) Yi = β1 + β2 X2i + β3 X3i + β4 X4i + β5 X5i + εi ,

(R) Yi = β1 + β2 X2i + β3 X3i + ui .
2 2
The above two models are estimated based on 65 observations and obtain R(U ) = 0.75 and R(R) = 0.7. At the
5% significance level, knowing F0.05 (2; 60) = 3.15, which model should we use?
H0 : (R) is more appropriate,
H1 : (U) is more appropriate.
F statistic:
2 2
(R(U ) − R(R) )/j (0.75 − 0.7)/2
F = 2 )/(n − p − 1) = = 6.
(1 − R(U ) (1 − 0.75)/(65 − 4 − 1)

Since F > F0.05 (2; 60), we reject H0 . Therefore, (U) is more appropriate.
16. Consider a data set on employee salaries including the following variables:
X2 : education level (number of years of schooling),
X3 : age,
X4 : experience (number of months worked),
X5 : starting salary (USD),
Y : current salary (USD),
G: gender (1 for male and 0 for female),
L: job position (1: employee, 2: team leader/foreman, 3: manager),
For the job position variable, using the dummy variable technique, we set:
( (
1 : manager, 1 : team leader/team leader/foreman,
L1 = L2 =
0 : other, 0 : other,

Estimating the model Y = β1 + β2 X2 + β3 X3 + β4 X4 + β5 X5 + β6 G + β7 X2 G + β8 L1 + β9 L2 + ε, we obtain

Source | SS df MS Number o f obs = 39
=============+================================== F(8 , 30) = ?
Model | Prob > F =
Residual | R=s q u a r e d = ?
=============+================================== Adj R=s q u a r e d = 0.875232
Total | Root MSE = 5676.253

==============================================================================
Y | Coefficient Std . e r r . t P>| t | [95% c o n f . i n t e r v a l ]
=============+================================================================
X2 | 4726.227 ? 2.236999
X3 | = 277.3656 260.2222 = 1.065880
X4 | 10.32685 26.29294 0.392761
X5 | 2.068516 0.409791 5.047738
G | 23483.66 18167.30 1.292634
c . X2#c .G | ? 2013.930 = 1.962362
L1 | = 36568.83 17644.68 = 2.072513
L2 | = 4389.050 2444.130 ?
_cons | = 21795.14 17345.88 = 1.256502
==============================================================================
(a) Calculate the coefficient of determination.
n−1
Adj R-squared: Ra2 = 1 − (1 − R2 ) n−p−1 .
Hence R2 = 1 − (1 − Ra2 ) n−p−1 39−8−1
n−1 = 1 − (1 − 0.875232) 39−1 = 0.9015.
(b) Calculate the F statistic.
(n−p−1)R2 (39−8−1)0.9015
F = p(1−R2 ) = 8(1−0.9015) = 34.3211.
(c) Calculate the standard error of β2 .
4726.227/2.236999 = 2112.7533.
(d) Calculate the point estimate of coefficient of X2 G.
2013.930 × (−1.962362) = −3952.0597.
(e) Test the hypothesis H0 : β9 ≥ −4000 against H1 : β9 < −4000.
t = −4389.050+4000
2444.130 = −0.16.
Since t > −t0.05 (39 − 8 − 1) = −1.7, we cannot reject H0 .
(f) What is economic interpretation of regression coefficients β5 , β7 , β8 , β9 ?
If starting salary increases by 1 USD, then the current salary increases by β5 USD.
β7 is the difference of current salary increases of males and females when the years of schooling increases
by 1 year.
Current salary of managers is greater than that of employees β8 USD.
Current salary of team leaders/foremen is greater than that of employees β9 USD.