ECN220 Project
Spring Semester 2022/23
The submission deadline for all components of the Applied Portfolio is Monday 15 May
by 12.00 noon. All submissions must be uploaded to the L2 Applied Portfolio Blackboard
organisation.
The document should be no longer than 6 pages (plus a maximum of 2 pages of references).
Use the “ECN Turnitin Submission Template.doc”" available on Blackboard/Turnitin. Failure
to attach the correct coversheet and include your student registration number will incur a 5%
deduction of your assignment mark which will be applied before any late submission penalties
have been imposed.
The module does not have dedicated workshops or lectures for the project, but there will
be signposts throughout the semester to support the task. The project should demonstrate
applications of concepts and techniques learned throughout the module; you should also be able
to provide a good interpretation of the results.
1
The behaviour of interest rates
The aim of the project is to study the behaviour of various interest rates across recessions
and financial crises in the United States. You will organize, graph and interpret the data. You
will also use various econometric tools to understand the behaviour of interest rates during the
business cycle.
1 Data cleaning and preliminary analysis
Log on to the Federal Reserve Economic Data website (FRED)
https://research.stlouisfed.org/fred2/categories
Download the following interest rates (feel free to download more interest rates of your
choice). Use monthly frequency from the starting observation of the series to the last observation
of 2022.1
1. Moody’s Seasoned Aaa Corporate Bond Yield (AAA)
2. Moody’s Seasoned Baa Corporate Bond Yield (BAA)
3. 3-Month Treasury Bill: Secondary Market Rate (TBILL3)
4. Effective Federal Funds Rate (FFR)
Load all the data in a single Excel file (or Stata file) and graph the series. Make sure that the
x− and y− axes are correctly labelled. Provide a table with some basic descriptive statistics:
minimum and maximum value, standard deviation, mean. Compute the correlation coefficient
across the series over the entire sample. Please report this.
Looking at the graph, provide a brief (about 500 words) comment on the dynamics of the
various interest rates. (Hint: use the economic theories studied in the module. You do not have
to limit yourself to the lecture on bonds. Other lectures are also useful to explain the data. You
can also draw on theories from other modules).
2 Empirical section
This section looks at the relationship between the interest rate spread and macroeconomic
variables.
1. Start by setting up the database.
1Data for the corporate bonds start in Jan-1919. The other two series start in 1954.
2
(a) Using the data collected in Section 1, generate the following corporate bond spread:
SPt = BAAt −AAAt
Where SPt is the spread at time t, BAAt and AAAt are the Baa Corporate Bond
Yield and the Aaa Corporate Bond Yield collected in Section 1.
(b) From the FRED download the Civilian Unemployment Rate (series ID UNRATE).
The unemployment rate should be at a monthly frequency, seasonally adjusted and
for the period 1954:01-2022:12.
(c) Generate dummy variable, D2007, which takes the value of 1 during the financial
crisis (December 2007 - June 2009), 0 otherwise.
(d) From the FRED download the Industrial Production: Total Index, Index 2017=100,
Monthly, Seasonally Adjusted (series ID INDPRO). The data should start from Jan-
uary 1919 to December 2022. Create the business cycle following the procedure of
Hamilton (2017). The procedure is the following:
yt+h = β0 + β1yt + β2yt−1 + β3yt−2 + β4yt−3 + vt+h
Where yt+h is the log of industrial production h periods ahead, β0,1,2,3,4 are coefficients
and vt+h is the error term. The stationary, or cyclical, component is then obtained
from the residuals
vˆt+h = yt+h − βˆ0 − βˆ1yt − βˆ2yt−1 − βˆ3yt−2 − βˆ4yt−3
The procedure can be implemented in Stata using the command HAMILTONFIL-
TER.2 The procedure will return the trend and the cycle, you only need the cycle.
Rescale the cycle by multiplying it by 100. Plot this variable with the SPt created
above.
(e) From the FRED website download the Consumer Price Index for All Urban Con-
sumers: All Items in U.S. City Average, Index 1982-1984=100, Monthly, Seasonally
Adjusted (series ID CPIAUCSL). The series starts in January 1947. Calculate the
annual inflation rate as:
πt = ln
(
Pt
Pt−12
)
× 100
Where πt is the inflation at time t, and Pt is the price level at time t.
2. Provide a table with the descriptive statistics of the variables you have just created (Min,
Max, St.Dev., Mean, number of observations).
3. Provide a correlation matrix among the variables described above. Make sure that the
start date is the same for all the variables to make the correlations comparable.
2You may have to install the package. In Stata type “ssc install HAMILTONFILTER”
3
4. Run the following regression:
SPt = α0 + α1yt + α2D2007 + α3FFRt + α4πt + ϵt
Where SPt is the spread, yt is the business cycle calculated above at time t, D2007 is the
dummy capturing the crisis, FFRt is the monetary policy at time t as in Section 1, πt is
the inflation rate at time t, ϵt is the error term.
5. Comment on the results obtained (max 1000 words).
(a) What is the economic significance? Do your results support the theory?
(b) Are the results econometrically sound? What are the limitations of this econometric
specification?
(c) What other method could you have used to overcome the econometric limitations?
6. Plot the SPt for the period January 2018 onwards. You should see two big movements in
the spread. Using the data downloaded above and the economic theories studied through-
out the semester provide an explanation for the behaviour of the spread. (Max 500 words;
you can use graphs and/or stats to support your analysis.)