Econmetrics - EC4061: t t t−1 0 t t−1 t 2 ε
Econmetrics - EC4061: t t t−1 0 t t−1 t 2 ε
Econmetrics - EC4061: t t t−1 0 t t−1 t 2 ε
Question 1
Assume that Yt is an ARM A (1, 1) time series;
Yt − φYt−1 = φ0 + εt − θεt−1
where εt is an i.i.d. (0, σε2 ) sequence.
1. State the conditions when;
(a) Yt is a stationary time series
(b) Yt is an invertible time series
2. Suppose that ARM A equations have a stationary solution Yt .
(a) Find E(Yt ) and V ar(Yt ). Note: For V ar(Yt ), assume for simplicity that
φ0 = 0
(b) Find auto covariance and autocorrelation function ρk , k ≥ 1. Assume
φ0 = 0
Question 2
Assume that Yt is a Moving Average M A(q) time series process given by:
Yt = εt − θ1 εt−1 − ... − θq εt−q
where εt is i.i.d. (0, σε2 ) sequence.
1. Show that E(Yt ) = 0
2. Show that V ar(Yt ) = σε2 (1 + θ12 + ... + θq2 )
Question 3
Consider the following rst-order vector autoregression (VAR) model in struc-
tural form.
1
c) The standard VAR can be estimated using ordinary least square (OLS)
or seemingly unrelated regression (SUR). State the conditions under which each
of these methods should be considered.
d) The standard form VAR can be estimated, but cannot be used to recover
the primitive VAR estimates. State why and provide condition(s) under which
this can be resolved.
Question 4
Consider the following rst-order vector autoregression (VAR) model in struc-
tural form.
Question 5
Consider the following rst-order vector autoregression (VAR) model in struc-
tural form.
Question 6
Consider the autoregressive distributed lag, ARDL, model given by