# ECON3031 Applied Econometrics

ECON3031 Applied Econometrics 2019-20
ASSESSED COURSEWORK 1
Answer all parts of the Assessed Coursework. You need to submit an electronic copy of your
answers via the designated turnitin link on Blackboard by Monday, 11th November 2019 at
23:59. Make sure that your answers do not exceed 10 A4 pages in total. Make sure you
explain your answers: donâ€™t just copy EViews plots and output. Please note that for
submissions received after the 11th November 2019, standard late submission penalties will
apply (see Blackboard for details).
The EViews workfile ukpi.wf1 (available on Blackboard) contains monthly data on price
indices for the UK and the Euro Area for the period January 1996 to December 2009.
a) From the monthly UK and Euro price indices (UK and EURO respectively), generate the
annual percentage rates (i.e. monthly series of % inflation rates over the past 12 months)
of price inflation, labelled UKINFL and EUROINFL respectively. Show which
transformation(s) you used to obtain the new variables.
b) Plot all four of the time series UK, EURO, UKINFL and EUROINFL, and obtain their
autocorrelation functions (ACFs). What patterns do you observe? Interpret the Qstatistics for each ACF.
c) Test for a unit root in each of the time series UK, EURO, UKINFL and EUROINFL using
the Augmented Dickey-Fuller (ADF) test. In each case, specify how you choose the lag
length and whether to include an intercept/linear trend in the ADF regression.
d) What are the orders of integration of UK, EURO, UKINFL and EUROINFL? Justify your
answers. Why is the order of integration important for cointegration analysis?
e) Estimate a linear regression between UK and EURO. Are UK and EURO cointegrated, or
f) Specify and estimate a bivariate Vector Autoregression (VAR) model for the two price
indices. In light of your answer to part d) above, justify your specification, including the
choice of lag length.
g) Specify and estimate an appropriate Autoregressive Distributed Lag (ARDL(p,q)) model
for UKINFL. Justify your choice of lags. Are the residuals serially correlated
(autocorrelated)? Explain why/why not.
h) For your chosen model for UKINFL from part g), obtain the ACF for the squared
residuals, and interpret your findings.
i) Using in each case your chosen specification from part g), apply Lagrange Multiplier
(LM) tests for ARCH effects in each of UKINFL and EUROINFL. What do you conclude
in each case?
j) Now estimate GARCH(1,1) models for the conditional variances of UKINFL and
EUROINFL. Interpret your estimates in the light of your answer to part i) above.
JMP