Econometrics I

Exam file
Spring 2022

Instructions: The exam file is individual work and is open-book. Please avoid
cheating. This means that helping your classmates or receiving help from them
or anybody is forbidden. Use your own reasoning; copying and pasting
definitions is not allowed. If you do not understand an exercise, contact me
and I can help you.

Interview:
Being honest in your work is rewarded in the interview.
You must attend your interview on your chosen day and send me your hand-
written solution (with ID and name on all pages) 10 minutes before your
interview. If you want me to consider your exam file, you must attend the
interview. You can answer in either Spanish or English.

SECTION I: Introductory part of the course. If you want to improve your grade of the first
exam, you can answer this section. If not, you can leave it blank and proceed with the other
sections (30%). -> 3 Questions (i.e. exercises)

1. Here below you find a dataset. The description of the columns is as follows:
“cve_inegi” is a code assigned from INEGI to each individual in the dataset.
“municipali~D is an ID assigned to each individual in the dataset. “State” and
“municipality” explain themselves. The last two columns from left to right represent
the Log of population (logpop) and the GDP per capita (gdppc) respectively.

a. Explain in detail, which type of data are these here below.

b. Suppose you want to use these data to build up an econometric model.
Please explain the model that you can build up using these data and provide
a plausible theory for your model.
c. According to the data type, you have explained in point a), is it true that the
variation that you have in this dataset is the best you can have. Explain why.
d. Using the dataset, provide a dispersion diagram and explain whether the
residuals are homogenous across observations.

2. Now, suppose you have already specified your model from exercise 1. According to
the Dependent Variable and Control Variable that you have selected, explain
whether with the data you have could be possible to identify a causal effect of your
chosen Control Variable on the Dependent Variable.
 3. Do you expect the error terms of the individuals under investigation to be
correlated?

SECTION II: Middle part of the course (25%+ 25% = 50%). -> 3 Questions (i.e. exercises)

During the Covid-19 there have been several challenges in education. One of those
challenges is to identify whether students have learnt the material in sufficient depth. For
this exercise, you need the provided Dataset in excel format
(Database_Exam_Education_Covid-19) sent to you together with this exam file.

Suppose you are working as an economist for the Ministry of Education. You have a group
of students whose size consists of 60 people participating in course A. You are asked to
investigate the determinants of the student’s final grade in that course. The first thing you
need to do is to set up an econometric model to give an answer to your boss.
In the provided data set in the Excel file, you have the following list of variables:

Days of attendance (from Day 1 up to Day 24): Binary variable being equal to 1 if the
student attends class on a specific day and 0 if not.
Attendance_1: Number of days the student attended the course.
Attendance_2: Binary variable being equal to 1 if the student attends the Review Session
and 0 if not.
Job_hours: Number of hours that a student works during a week.
Grade_1: Grade of the student in course A (a grade higher than 7.5 is pass)
Grade_2: Grade of the student in course A
Open_book: Binary variable being equal to 1 if the student has the option to present an
open-book exam and 0 if not.
Days to interview: number of days that the student has from the delivery date of the exam
file up to the interview.
Residual_1: Estimate of the error term, which is unique to each individual.
Residual_2: Second estimate of the error term, which is unique to each individual.
Study hours in semester: Number of hours that the student has learnt at home during the
semester

1. Set up an econometric model useful to investigate the determinants of the student grade.
a. Observe the dataset in the Excel file. Is it possible to use all variables there as
controls?
b. You have two estimates of the error (i.e. two columns for the residual). As you can
see, both columns are different. What is a plausible explanation for the difference
that we observe in both residuals?

2. Suppose that in your data set you have a measure of cheating for each student in the
open-book exam. Let us say that the variable job hours is 𝑋4𝑖 and you want to test the
hypothesis: 𝐻0 : 𝛽4 = 0. Given the presence of the cheating variable, is it more likely to
reject the null or not? What type of variation do you expect to observe in the dependent
variable?

3. Use the model you proposed in point 1 of this section. Suppose now that the residuals
you have are as in the column Residual_2. You are given the first and second moment of
the distribution of these residuals such that you have: Mean= -0.05 and Variance= 0.93 Can
we say that these residuals are normally distributed? Why? And if they are, can we test the
significance of the control variables either by using the Significance Testing-Method or the
Confidence Interval Method? Explain your answer.

SECTION III: Last part of the course. (20%) -> 1 Question (i.e. exercise) 5% and interview
15%

1. Now, your boss sends you to the World Bank in Washington D.C. to attend a very
important meeting with other economists. The objective of this meeting is to present
estimates of the effect of the hybrid study system (due to the Covid-19 pandemic) on the
knowledge quality of students. Propose one way of measuring the hybrid system variable
and include it in your model. Do you think that the hybrid study system variable will
contribute to maintain the normality assumption of the error term in your model? Please
explain.