QUESTIONS
At the beginning of your report, briefly explain how you prepared your data files, especially
demonstrating their originality.
Answer the following questions using the 188 quarterly observations (Q3/1972 ‒ Q2/2019) on
Q1. Visual Checking
Plot each of the six variables against time, and obtain ACF and PACF plots. Comment on the
plots, ACFs and PACFs, with regard to the trend and stationarity/non‐stationarityof the series.
Q2. Order of Integration
• Determine the orders of integration for the variables using the Augmented
Dickey‐Fuller (ADF) test procedure at 5%. For the tests, set the maximum lag order to
8 and select “test down” from it using the AIC to determine the optimum number of
lags. Include constant, and constant and trend for the tests as necessary.
• Make sure you test for I(1) vs I(0) as well as I(2) vs I(1),
Q3 Engle-Granger cointegration test
Test for cointegration test among the six variables LTHPI, Rrate, LRGDIPC, LUN, LCPI and
LHSTOCKPC using the Engle-Granger test.
Q4 Long run relationship
Whether or not you found cointegration using the Engle-Granger test estimate the long-run
relationship (assuming it exists) between those six variables using OLS, including a constant in
the relationship.
Note : The Engle-Granger cointegration test has low power (the probability of rejecting the null of “no
cointegration” when the null is false is low). Other tests are available but their implementation
is complex. One of the them is the Johansen test. Using the Johansen test I found cointegration
among the variables exist.
Q5 Economic Interpretation of long run relationship
• Explain why you would expect a long run relationship between those six variables.
• Interpret the estimated coefficients of the explanatory variable.
• Are the signs and magnitudes of the estimated coefficients according to intuition?
• Comment on the cointegration relationship in light of the economic events that
happened during the sample period, paying attention to the signs of the
coefficients.
Q6 Estimation of the long-run relationship using DOLS
• Estimate the long-run relationship with DOLS using 4 leads and 4 lags. Make sure
you use the HAC standard errors.
• Are the signs and magnitudes of the estimated coefficients according to intuition?
• Are the coefficients significant?
• Assuming cointegration exists are the statistical tests valid?
Note: It is awkward to create leads in Gretl. It requires two steps. For example, if we want to
create a 2-lead in the first difference of LUN, assumed to have generated as dLUN: 5dun2=dLUN(2)
dunplus2=dun2
You have to repeat these steps for each explanatory variable and each lead.
Q7. Short-run Dynamic Relationship: ARDL
Estimate a short‐run dynamic model using an ARDL model using the first difference of the six
variables. Explain how you chose the number of lags both for the lagged dependent variable
and lagged explanatory variables.
Q8. Short-run Dynamic Relationship: ECM
Estimate a short‐run dynamic model using an ECM model as per page 34 of the LEC 6 lecture
notes. Is the error correction term significant? Does it have the appropriate sign for
convergence to equilibrium.
For the hypothesis tests, clearly state the test equation (including the number of lagged terms),
test statistic, significance level, p-value or critical value (from the table), decision and conclusion.If
the same test procedure is applied to multiple cases, description of all the test steps is required
only for the first of the tests based on the same procedure.
Copy and paste a relevant part of the Gretl output to support your answer to each question. Note,
however, that copy‐and‐pasted Gretl outputs can only be complementary to your answer and they
alone cannot be an answer.