Final Assessment Information
Contents
Preliminary information………………………………………………………………………………………………………………..2
Task 1: Things you have do using the STATA statistical Software …………………………………………………….3
Task 2: Questions and explanations that you need to put in the Microsoft word document…………………….7
How will you be graded………………………………………………………………………………………………………………..9
2/ 9
Preliminary information
Your final assessment is worth 15% of the grade.
The assessment is due by 11:59 pm on Friday November 20. Monday December 7
For the final assessment, you will use the STATA statistical software, course material posted on
OWL, and other online resources to execute a series of commands and answer some related
questions.
Can I work in a team of by myself?
• You can complete this final assessment by yourself or in a team.
• If you do decide to complete the assessment in a team, please note that teams can be
different or the same as the team that you are in for project 1 and project 2.
• If you work in a team, one final assessment should be submitted for the team. Please
list the team members on the first page.
What do you hand in?
Please submit these files on OWL.
1. Word document
2. Stata Do file
What should you keep in case I ask for it?
3. Excel dataset
4. Stata dataset
5. Log file
Academic integrity
**WARNNG** Any instances of where non-compliance with academic integrity may result in a zero
grade in the course and a notation on your transcript.
Few things to note:
• Before a final grade is assigned for your assessment, one or more member may be selected at
random for an interview with the instructor. In this interview, this team member will be asked
questions about the assessment. The response of this team member may help to inform the
final grade of the team.
3/ 9
Task 1: Things you have do using the STATA statistical Software
**************
Submission requirements:
• Stata do-files
***********
Complete in the order shown
Open the STATA statistical program
For how to access STATA go here: Scotty Server Information to access STATA
1) Do file and Log file
Create a STATA do file
https://sociology.fas.harvard.edu/files/sociology/files/creating_a_do_file.pdf

2) Create a STATA log file
https://sociology.fas.harvard.edu/files/sociology/files/creating_a_log_file.pdf

The log file must be referenced in the do file. The log file will keep track of all the commands
once you execute the commands.
3) Go to the following website:
https://www150.statcan.gc.ca/t1/tbl1/en/tv.action?pid=3610010401
You want to download data from 2000Q1 to 2020Q2. Make the changes in the reference period
tab as shown below.
4/ 9
Then click “Apply”
Thereafter download the data.
Your interest will be in three series:
Gross domestic product at market prices
Household final consumption expenditure
Business gross fixed capital formation
4) The series will be download in rows, transpose the data so that they in columns.
5) Covert these series to seasonally adjusted at quarterly rates by dividing them by four.
6) Rename your series as follows in the excel file
gdp = Gross domestic product at market prices
con = Household final consumption expenditure
inv = Business gross fixed capital formation
7) Import your dataset with the four series into stata. You should have a file that looks like
gdp con inv
xxx xxx xxx
xxx xxx xxx
. . .
. . .
5/ 9
For help, see:


8) In your STATA do file type the command to create a new variable called qtr.
This variable will be your time variable.
Type the commands to make the values of the variable such that the starting value is 2000Q1 and
that ending value is 2020Q2
execute /run the do file
make/set the variable qtr to be your time variable.
For help:
http://www.learneconometrics.com/class/5263/notes/Defining%20Time.pdf

9) In the do file enter commands to graph these variables gdp, con, inv. Please graph each of
these series separately.
10) Report the summary statistics for these variables gdp, con, inv. Enter the commands in the do
file.
11) Transform your existing variables by taking the natural logarithms of the variables. Enter the
commands in the do file to generate a new variable.
The new variables should therefore be
lngdp = ln(gdp)
lncon = ln(con)
lninv = ln(inv)
see help here: https://www.youtube.com/watch?v=E_wCh0rf4p8
12) Use the Augmented Dickey Fuller (ADF) test to check unit roots of the levels of the variables
lngdp, lncon, & lninv. Do so for the period 2000q1 to 2019q4 Hint: include the command
if tin(2000q1, 2019q4)
Carryout and report three versions of the test:
6/ 9
(1) without constant(intercept) and trend
(2) with constant (intercept) and no trend
(3) with constant(intercept) and trend
13) Use the Augmented Dickey Fuller (ADF) test to check unit roots of the first difference of
the variables lngdp, lncon, & lninv. Do so for the period 2000q1 to 2019q4 Hint: include
the command if tin(2000q1, 2019q4)
Carryout and report three versions of the test:
(1) without constant(intercept) and trend
(2) with constant (intercept) and no trend
(3) with constant(intercept) and trend
14) Your friend Emily runs a regression where lngdp is the dependent variable with lncon and
lninv are the independent variables. Do so for the period 2000q1 to 2019q4
15) Your friend Jupiter runs a regression where D.lngdp is the dependent variable with D.lncon
and D.lninv are the independent variables. He does so for the period 2000q1 to 2019q4
16) Input and run the following commands: vecrank lngdp lncon if tin(2000q1, 2019q4), lags(2)
trend(trend) max trace
17) Consider the variable lngdp. Generate ACF and PACF for the period 2000q1, 2019q4
See here for help.
18) Consider the variable D.lngdp. Generate ACF and PACF for the period 2000q1, 2019q4
See here for help.
19) Estimate an ARIMA model for d.lngdp or d.lncons or d.lninv using the Box-Jenkins
methodology for the period 2000q1 to 2019q4. Click links: link 1 and link 2 for help.
Also see: https://www.youtube.com/watch?v=8xt4q7KHfBs
20) Estimate the following basic VAR by entering the commands: var d.lngdp d.lncon d.lninv if
tin(2000q1, 2019q4), lags(2)
https://www.stata.com/manuals13/tsvargranger.pdf
21) Conduct a granger causality test using the basic VAR that you estimated previously
https://www.stata.com/manuals13/tsvargranger.pdf
7/ 9
Task 2: Questions and explanations that you need to put in the Microsoft word document.
****
Submission requirements.
Explanations, graphs, and tables as indicated.
Explain in your own words. Copy and pasting from the intent will give you zero marks. Any
source you use must be cited at the any of your word document.
The instructor may conduct an interview to check veracity and understanding of the explanations
given. This will be done via zoom.
****
1) Explain why we seasonally adjust a time series.
2) Explain the difference between trend, cycle, seasonal, and irregular components of a time
series.
3) What is the difference between gross fixed capital formation and gross capital formation?
4) Consider the graphs from STATA for series gdp, con, inv. Put these graphs in the word
document. Discuss what you notice about these graphs. Here is some help
5) Put the descriptive tables in the word document. Discuss the descriptive statistics
associated with gdp, con, inv. Here is some help.
6) Explain why we often take the natural logarithm of series before we carryout analysis on
them.
7) Explain the difference between trend stationary and difference stationary time series.
8) In theory, first differencing a trend stationary series introduces a unit root. In practice,
however, researchers often decide to take the first difference of a series even though test
results might suggest that it is trend stationary. Explain how one could justify such a
decision.
9) Discuss how one would go about choosing the number of lags to include in testing for
unit roots
8/ 9
10) Put all the tables of the different versions of the Augmented Dickey Fuller test on the
levels for lngdp, lncon, & lninv in the word document. Interpret your results.
11) Put all the tables of the different versions of the Augmented Dickey Fuller test on the first
difference for lngdp, lncon, & lninv in the word document. Interpret your results.
12) Explain why it might be useful to report three different versions of the test, namely
without constant(intercept) and trend, with constant(intercept) and trend, and with
constant(intercept) and no trend.
13) Consider the regressions done by Emily and Jupiter. Report these regressions in the
work document. Which regression do you prefer? Discuss.
14) Explain what cointegration means.
15) Explain what value there is in knowing that a set of series are cointegrated.
16) Explain the null and alternative hypotheses associated with the Johansen cointegration
test
17) Explain thoroughly the results that you obtained from:
vecrank lngdp lncon if tin(2000q1, 2019q4), lags(2) trend(trend) max trace
18) Explain what error correction model means. Do the results from vecrank lngdp lncon if
tin(2000q1, 2019q4), lags(2) trend(trend) max trace indicate that we can use such a
model.
19) In an error correction model, what does the coefficient on the error correction term tell
us?
20) How do ACF and PACF help us in ARIMA modelling?
21) Put the graphs of the ACF and PACF that you created in the word document.
22) Explain how you used the Box-Jenkins methodology to estimate an ARIMA model for
d.lngdp or d.lncons or d.lninv.
23) Discuss the adequacy of your ARIMA model of d.lngdp or d.lncons or d.lninv
24) Explain what a VAR model is, and the steps involved in estimating such a model.
25) Explain the granger causality results that you obtained form the VAR model that you
estimated.
26) Explain what impulse response functions are in the context of VAR models.
27) Explain
a. Static forecast versus dynamic forecast
9/ 9
b. In sample forecast versus out of sample forecast
How will you be graded.
You will be awarded a mark out of 15% based on the following:
Grade Descriptor
90-100 Brilliant. Zero or near errors. Very strong evidence of complete mastery in
the analysis and reporting of the results.
80-89 Outstanding. Some errors are present but strong evidence of comprehensive
understanding in the analysis and reporting of the results.
70-79 Very Good. More errors are present, but there is evidence of very good
understanding in the analysis and reporting of the results.
60-69 Good. A good effort with weaker evidence of a very good understanding
shown the analysis, reporting, and explanation of the results.
50-59 Average.
0-49 No submission/thing piece of work/plagia