WILFRID LAURIER UNIVERSITY

WATERLOO | BRANTFORD | KITCHENER | TORONTO

TERM: Fall 2024

COURSE Code: MGT604
COURSE TITLE: Applied Econometrics Name:
INSTRUCTOR(S): Jean Eid
Number of Pages: 9
wlu I.D.#:
Length of Examination: Two Hours
Aids allowed: Aids Allowed: Non-programmable calculator

The doors of the examination room will be opened approximately 10 Candidates must not use or attempt to use any improper source of
minutes before the start of the examination. Candidates will be per- information. No candidates for an examination may bring into the
mitted to enter the examination room quietly up to one half hour after examination room any books, notes or other material containing infor-
the scheduled start of the exam. Candidates arriving late will not be mation pertaining to the examination unless the examiner has given
allowed any extra time. instructions that such material will be allowed and this instruction is
specified on the examination paper. Any item brought into the exami-
Candidates must not begin the examination or attempt to read the nation room is subject to inspection.
examination questions until instructed to do so.
No briefcases, backpacks or other bags and carriers may be brought to
Candidates once having entered, may not leave the exam room before the desk site where the candidate is writing the examination. These
completing and submitting the exam unless accompanied by a Proctor. bags should be left outside the examination room. If books, notes etc.
Candidates are not permitted to submit their examination and leave the cannot be left outside the examination room, they must be put at the
examination room until 1 hour after the examination has begun, and front of the examination room in a place designated by the proctor
in no case before their attendance has been taken. In no case may a before a candidate takes a seat. Candidates are advised not to bring
candidate leave the room temporarily, for any reason, until 30 minutes valuables to the examination room.
after the start of the examination. In order that remaining candidates
are not disrupted, candidates must remain seated and may not leave Unless specified by the professor, no calculator, electronic or commu-
the examination room during the last 15 minutes of the examination nication devices will be allowed in the examination room, including cell
session. phones, smartphones, etc. Cell phones will be taken away if found and
an Irregularity notice will be filed with the Integrity Office. Allowed
At the close of the examination period, candidates must stop writing aids must be indicated on the examination paper. If non programmable
immediately. The Presiding Officer may seize the papers of candidates calculators are permitted, only those without lids, authorized by the
who fail to observe this requirement, and a penalty may be imposed instructor, will be allowed. It is the candidate’s responsibility to ascer-
at the discretion of the instructor. Candidates must submit all their tain whether the use of calculators is permitted, and, if it is, whether
work, according to the instructions of the Presiding Officer, including any restrictions are imposed on the types of calculators that may be
all materials and a copy of the examination paper with their name and brought to the examination. No pencil cases are allowed on the desks.
student ID number written on it. Unused examination booklets may
not be taken from the examination room. Translation dictionaries (e.g. English French) or other dictionaries,
(thesaurus, definitions, technical) are not allowed unless specified by
A candidate who leaves before the examination is over must hand in the instructor and indicated on the examination paper. Electronic dic-
all completed and attempted work, notes made during the exam, and a tionaries are never allowed.
copy of the examination paper with their name and student ID number
on it. Except for bottled water (with label removed), no food or drink is al-
lowed in the examination room. Candidates with health problems that
Talk or any form of communication between candidates is absolutely warrant relaxation of this regulation should provide medical documen-
forbidden. No information of any kind is to be written on the question tation to the presiding officer prior to the beginning of the examination.
paper or on scrap paper for the purpose of assisting other candidates. Such students should restrict themselves to those items and packaging
Responses to questions must not be done in an exaggerated way or in that will least distract other examinees.
a manner that will involve transmission of information to others.
Candidates are expected to write their examinations in an honest and
Candidates must remain seated during the examination period. A straightforward manner. Where there are reasonable grounds for be-
candidate needing to speak to the proctor (e.g. to ask for additional lieving a violation of exam protocol has occurred, the candidate will be
supplies or to request permission to leave the examination room for subject to the disciplinary procedures and sanctions according to the
any reason) should so indicate by raising their hand. University Calendar.

Questions concerning possible errors, ambiguities or omissions in the Only currently registered students will be permitted to write the final
examination paper must be directed to the proctor who will investi- exam.
gate them through the proper channels. The proctor is not permitted
to answer questions other than those concerning the examination paper. Examinations conducted at Wilfrid Laurier University will be bound by
WLU regulations, regardless of where the candidate is registered.

Approved by Senate ( Oct 27/2003) Updated Sept 2019

75 University Avenue West, Waterloo, Ontario N2L3C5 LazaridisSchool.ca

T: (519) 884-0710
This is the final exam for Applied Econometrics (MGT604 ). Please answer the questions as clearly and in as
much detail as possible.
This examination consists of 9 (including 2 extra pages for your work if needed) pages with 4 questions for a
total of 150 points.
Please bring any discrepancy to the attention of an invigilator. The number in parentheses at the start of each
question is the number of points the question is worth.
Answer the 4 questions.

For Grader’s use:

Question Points Score
Q1 0
Q2 50
Q3 50
Q4 50
Total: 150
 Multiple Choice Questions
Question 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0 points
Answer all the following multiple choice questions. Please circle the letter of your answer.
Question 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50 points
Interpretation and hypothesis testing of regression coefficients
Suppose you have a cross sectional data set on student attendance in an econometrics class with the
following variables

Var Label

attend classes attended out of 32

termGPA GPA for term
priGPA cumulative GPA prior to term
ACT American College Testing (ACT) score (max=36)
final final exam score
atndrte percent classes attended
hwrte percent homework turned in
first =1 if in first year
second =1 if in second year
missed number of classes missed
stndfnl (final - mean)/sd

and the following summary statistics

Variable Obs Mean Std. dev. Min Max

attend 678 26.20501 5.351804 3 32
termGPA 678 2.608673 0.7239593 0.321 4
priGPA 678 2.588521 0.5443898 0.857 3.93
ACT 678 22.50442 3.494137 13 32
final 678 25.89823 4.711151 10 39
atndrte 678 81.89067 16.72439 9.375 100
hwrte 678 88.09524
first 678 0.2330383 0.4230786 0 1
second 678 0.5752212 0.4946743 0 1
missed 678 5.794985 5.351804 0 29
stndfnl 678 0.03114077 0.9897376 -3.308824 2.783613

Question 2 continues Final(MGT604 –Fall–2024) Page 1 of 9

on next page
 a: You want to know whether attending the class results in a better grade for students on the final exam.
In you issue the following commands
R Code for Q 2- a

> reg2_1 <- lm(final~attend, data=attend)

> reg2_2 <- lm(final~attend+ACT, data=attend)
> reg2_3 <- lm(final~attend+ACT+ hwrte, data=attend)
> reg2_4 <- lm(final~attend+ACT+ hwrte+ first , data=attend)
> reg2_5 <- lm(final~attend+ACT+ hwrte+ first +second , data=attend)
> reg2_6 <- lm(final~attend+ACT+ hwrte+ first +second +priGPA, data=attend)

and we get the following from a huxreg command

(1) (2) (3) (4) (5) (6)
(Intercept) 22.767 9.452 8.675 8.896 9.313 9.452
(0.897) (1.459) (1.506) (1.543) (1.561) (1.531)
attend 0.119 0.172 0.124 0.126 0.127 0.031
(0.034) (0.031) (0.041) (0.041) (0.041) (0.044)
ACT 0.530 0.528 0.523 0.525 0.399
(0.048) (0.048) (0.049) (0.049) (0.053)
hwrte 0.024 0.023 0.023 0.021
(0.011) (0.011) (0.011) (0.011)
first -0.265 -0.817 -0.254
(0.402) (0.518) (0.519)
second -0.742 -0.821
(0.439) (0.431)
priGPA 2.055
(0.392)
N 678 678 672 672 672 672
2
R 0.018 0.169 0.175 0.176 0.179 0.212
2
adj. R 0.017 0.167 0.171 0.171 0.173 0.205
i. (2 points) How many students are in their third year and above
ii. (2 points) Explain why the number of observations goes down in the third regression
iii. (8 points) carefully interpret the coefficients of the last regression (6) (zero marks for positive or
negative relationship or association)
iv. (5 points) How would you explain the changes in the ACT and attend coefficients from the first
five regressions to the last
v. (3 points) Why would it be a bad idea to include attend and atndrte in the same regression
b: Concentrate on the last model (6):
i. (5 points) Which of the coefficients are individually statistically significant at the 5% level.

Question 2 continues Final(MGT604 –Fall–2024) Page 2 of 9

on next page
 2.5% 2.5% 5% 5%

−∞ −1.96 1.96 ∞ −∞ −1.65 1.65 ∞

10% 10%

−∞ −1.28 1.28 ∞

ii. (2 points) What do you think of the overall fit of the model? how much of the variation in the
dependent variable is explained by variation in the explanatory variables.
iii. (3 points) Test whether or not the explanatory variables simultaneously explain the dependent
variable.
iv. (6 points) We want to test whether the partial effect of ACT is the same as the priGPA. Here is
some code and output. Can you conduct the test given the information presented below? if
so, conduct the test, if not explain why.
R Code for Q 2- b.iv

> nossummary(reg2_6)

Question 2 continues Final(MGT604 –Fall–2024) Page 3 of 9

on next page
 R Output for Q 2- b.iv

Call:
lm(formula = final ~ attend + ACT + hwrte + first + second +
priGPA, data = attend)

Residuals:
Min 1Q Median 3Q Max
-14.0833 -2.6871 -0.1738 2.8526 10.9272

Coefficients:
Estimate Std. Error t value Pr(>|t|)
(Intercept) 9.45211 1.53067 6.175 1.15e-09
attend 0.03128 0.04408 0.710 0.4782
ACT 0.39891 0.05346 7.462 2.68e-13
hwrte 0.02111 0.01096 1.926 0.0545
first -0.25449 0.51873 -0.491 0.6239
second -0.82139 0.43062 -1.907 0.0569
priGPA 2.05513 0.39192 5.244 2.12e-07

Residual standard error: 4.215 on 665 degrees of freedom

(6 observations deleted due to missingness)
Multiple R-squared: 0.2116,^^IAdjusted R-squared: 0.2045
F-statistic: 29.75 on 6 and 665 DF, p-value: < 2.2e-16

R Code for Q 2- b.iv

> attend <- attend%>%

> mutate(ACTpriGPA=ACT*priGPA)
> nossummary(lm(final~attend+ACTpriGPA+hwrte +first +second , data=attend))

Question 2 continues Final(MGT604 –Fall–2024) Page 4 of 9

on next page
 R Output for Q 2- b.iv

Call:
lm(formula = final ~ attend + ACTpriGPA + hwrte + first + second,
data = attend)

Residuals:
Min 1Q Median 3Q Max
-14.131 -2.711 -0.101 2.938 11.091

Coefficients:
Estimate Std. Error t value Pr(>|t|)
(Intercept) 17.973527 1.020169 17.618 <2e-16
attend -0.005689 0.040646 -0.140 0.8887
ACTpriGPA 0.116598 0.009546 12.215 <2e-16
hwrte 0.019468 0.010934 1.780 0.0755
first -0.148984 0.514612 -0.290 0.7723
second -0.839521 0.430148 -1.952 0.0514

Residual standard error: 4.211 on 666 degrees of freedom

(6 observations deleted due to missingness)
Multiple R-squared: 0.2121,^^IAdjusted R-squared: 0.2062
F-statistic: 35.86 on 5 and 666 DF, p-value: < 2.2e-16

v. (7 points) Test whether the coefficient on hwrte is double that of attend by explaining the way
you would conduct the test. This is explaining the test and no code is needed. If you only give
code, you receive a mark of zero.
vi. (7 points) You want to test βACT = .5 & βpriGPA = .6. Explain the way you would conduct the
test. This is explaining the test and no code is needed. If you only give code, you receive
a mark of zero.

Question 3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50 points
Suppose you have data on school achievements. The data set is for California and has the following variables

Question 3 continues Final(MGT604 –Fall–2024) Page 5 of 9

on next page
 Var Label

county indicating county (factor)

testscr Average of two primary scores, math and reading
str Student teacher ratio
region 0 =Bay area, 1=Central, 2=Southern, 3=Superior
bay_area =1 if region is bay_area
central =1 if region is central
southern =1 if region is southern
superior =1 if region is superior
where the unit of observation is a school, and the dependent variable is average test score from a standardized
test(testscr). Student-Teacher Ratio is a continuous variable and is computed by dividing the total number
of students by the total number of teachers in a given school. In addition there is a region variable (region)
that is a categorical variable that describes the California Counties by Mental Health regions. There are
four regions plus Los Angeles, but for the purpose of our study we will absorb Los Angeles in the Southern
region.
a: (5 points) Explain in as much detail why the following regression does not make any sense
R Code for Q 3- a

> nossummary(lm(testscr~region, data=schools))

R Output for Q 3- a

Call:
lm(formula = testscr ~ region, data = schools)

Residuals:
Min 1Q Median 3Q Max
-49.903 -13.667 0.393 13.543 48.257

Coefficients:
Estimate Std. Error t value Pr(>|t|)
(Intercept) 658.4935 1.5620 421.564 < 2e-16
region -3.0409 0.8862 -3.432 0.00066

Residual standard error: 18.81 on 418 degrees of freedom

Multiple R-squared: 0.0274,^^IAdjusted R-squared: 0.02507
F-statistic: 11.78 on 1 and 418 DF, p-value: 0.0006598

b: Now suppose we do the following

R Code for Q 3- b

> nossummary(lm(testscr~central + southern+superior, data=schools))

Question 3 continues Final(MGT604 –Fall–2024) Page 6 of 9

on next page
 R Output for Q 3- b

Call:
lm(formula = testscr ~ central + southern + superior, data = schools)

Residuals:
Min 1Q Median 3Q Max
-42.789 -12.455 -0.657 13.774 48.996

Coefficients:
Estimate Std. Error t value Pr(>|t|)
(Intercept) 665.989 1.776 375.080 < 2e-16
central -18.365 2.408 -7.627 1.65e-13
southern -16.535 2.365 -6.993 1.08e-11
superior -9.147 2.705 -3.382 0.000787

Residual standard error: 17.67 on 416 degrees of freedom

Multiple R-squared: 0.1464,^^IAdjusted R-squared: 0.1402
F-statistic: 23.78 on 3 and 416 DF, p-value: 3.181e-14

i. (5 points) Interpret the coefficients From the regression above. Are there any differences between
the regions
ii. (5 points) What is the difference between the central and southern region in terms of test scores
c: We now issue the following commands
R Code for Q 3- c

> schools <- schools%>%

> mutate(strbay=str*bay_area)
> reg3_1 <- lm(testscr~str, data=schools)
> reg3_2 <- lm(testscr~str+county, data=schools)
> reg3_3 <- lm(testscr~str+bay_area+strbay, data=schools)
> reg3_4 <- lm(testscr~str+bay_area+strbay, data=schools)

(1) (2) (3) (4)

(Intercept) 698.933 *** 710.493 *** 664.972 *** 664.972 ***
(9.467) (17.998) (10.142) (10.142)
str -2.280 *** -1.101 * -0.729 -0.729
(0.480) (0.509) (0.508) (0.508)
bay_area 105.151 *** 105.151 ***
(22.366) (22.366)
strbay -4.768 *** -4.768 ***
(1.165) (1.165)
County Fixed effect No Yes No No
N 420 420 420 420
R2 0.051 0.407 0.178 0.178
adj. R2 0.049 0.336 0.172 0.172
*** p < 0.001; ** p < 0.01; * p < 0.05.
i. (2 points) What do the fixed effect in the second regression estimate?
ii. (3 points) Do the fixed effect matter ?

Question 3 continues Final(MGT604 –Fall–2024) Page 7 of 9

on next page
 iii. (5 points) Interpret the coefficient on the variable bay_area in model (3) . Does it make sense?
CODE
iv. (5 points) Interpret the coefficient on the variable bay_area in model (4) . Does it make sense?
v. (5 points) Interpret the coefficient on the variable strbay in model (4) . Does it make sense?
vi. (5 points) What is the total effect of one additional unit increase in Student-Teacher Ratio for
schools in the bay area, based on the results shown in the table?
vii. (5 points) What is the total effect of one additional unit increase in Student-Teacher Ratio for
schools that are not in the bay area , based on the results shown in the table?
viii. (5 points) Explain what we would be estimating if we include the fixed effects for the counties in
the last two regressions

Question 4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50 points
This question has two parts that are independent of each other
a: You have the following output from

dependent: S Coefficient Std. err. t-stat p>|t| [95% conf. interval]

(Intercept) -2.8571 0.7369 -3.88 0.000 -4.3047 -1.4095
exp1 0.5983 0.0515 11.61 0.000 0.4971 0.6995
exp2 0.0223 0.0121 1.85 0.065 -0.0014 0.046
exp3 0.1693 0.0217 7.81 0.000 0.1268 0.2119
exp4 -0.0676 0.4452 -0.15 0.879 -0.9422 0.807
R2 0.3065 SSR 4966.08
adj. R2 0.3011 SST 7160.41
sigma 3.0874 ESS 2194.33
F-stat 57.5527
N 526

where exp4 is a dummy variable

i. (5 points) Suppose you issue the following commands in

R Code for Q 4- a.i

> data <- data%>%

> mutate(exp11=exp1*10)
> lm(dep.var~exp11+exp2+exp3+exp4, data=data)

What is the effect of the change on the coefficients, fitted values and residuals

Question 4 continues Final(MGT604 –Fall–2024) Page 8 of 9

on next page
 ii. (5 points) Suppose you issue the following commands in
R Code for Q 4- a.ii

> data <- data%>%

> mutate(exp22=exp2/10, dep.var=wage)
> lm(dep.var~exp1+exp22+exp3+exp4, data=data)

What is the effect of the change on the coefficients, fitted values and residuals
iii. (5 points) Suppose you issue the following commands in
R Code for Q 4- a.iii

> data <- data%>%

> mutate(exp44=1-exp4)
> lm(dep.var~exp1+exp2+exp3+exp44, data=data)

What is the effect of the change on the coefficients, fitted values and residuals
iv. (10 points) Suppose you issue the following commands in
R Code for Q 4- a.iv

> lm(dep.var~exp11+exp22+exp3+exp44, data=data)

What is the effect of the change on the coefficients, fitted values and residuals
b: (25 points) In the output from below, fill out the missing values where you see the word calculate
(assume tdf,.975 = 1.96)

dependent: S Coefficient Std. err. t-stat p>|t| [95% conf. interval]

(Intercept) Calculate 310.4455 -1.53 0.127 -1083.5075 135.0056
exp1 -7.4318 Calculate -0.61 0.543 -31.3736 16.51
exp2 5.0587 0.9319 Calculate 0.000 3.2299 6.8876
exp3 59.3458 6.9931 8.49 0.000 Calculate Calculate
exp4 Calculate 3.0953 Calculate 0.000 9.9571 22.1064
exp5 185.2138 Calculate 4.74 0.000 Calculate Calculate
exp6 Calculate 0.1215 0.4 0.689 Calculate Calculate
R2 0.1835 SSR 124696613.9
adj. R2 0.1782 SST 152716168.22
sigma Calculate ESS 28019554.32
F-stat 34.7539
N 935

Extra Pages for your scrap work Final The End

MGT604 –Fall–2024