UNIVERSITY OF ZIMBABWE

BSc Honours in Statistics Level 2 HSTS203

TIME SERIES ANALYSIS

November/December 2018
Time : 2 hours

Candidates should attempt ALL questions in section A and TWO questions in section B.
Marks will be allocated as indicated

SECTION A (40 marks)

Candidates may attempt ALL questions being careful to number them A1 to A5.

A1. Let {at } be zero mean white noise. Show that the autocorrelation functions for the
following processes are the same:

(a) Zt = at + 41 at−1 . [2]

(b) Zt = at + 4at−1 . [2]

A2. Write down the equations for the following models:

(a) ARIMA(1,1,2). [4]

(b) ARIMA(1,1,0)(0,1,1)4. [4]

A3. Suppose {Yt } is an integrated process is given by:

Yt = 2Yt−1 − 1.25Yt−2 + 0.25Yt−3 + at

where {at } is a white noise process with E(at ) = 0 and V ar(at ) = σa2 . Show that Yt is
non stationary and hence find the order of integration. [4,1]

A4. Consider the following AR(2) process

1
Zt = Zt−1 − Zt−2 + at
2
(a) Show that the model is stationary. [3]

page 1 of 3
 HSTS203

(b) Showing all your working, deduce that the autocorrelation function of Zt is given
by:
  k2     
1 kπ 1 kπ
ρk = cos + sin for k = 0, ±1, ±2, . . . [8]
2 4 3 4

A5. Consider an AR(1) model given by:

Zt = φZt−1 + at

(a) Use the method of Least Squares Estimation to estimate the parameter φ. [6]
(b) Use the method of Maximum Likelihood Estimation to estimate the
parameter φ. [6]

SECTION B (60 marks)

Candidates may attempt TWO questions being careful to number them B6 to B8.

B6. (a) Given a general linear process

∞
X
Zt = ψj at−j = Ψ(B)at
j=0
P∞
where j=0 ψ 2 < ∞ and at is a white noise process with variance σa2 .
(i) Show that the process is stationary. [10]
(ii) Let γ(B) be the autocovariance generating function. Show that

γ(B) = σa2 ψ(B)ψ(B −1 ) [4]

.
(b) Let {Zt } be a general linear process expressed as:
∞  j
X 1
Zt = at−j
j=0
3

where {at } is a white noise process. Find the autocovariance generating func-
tion. [8]
(c) Let {γ(k)} be the autocovariance function of a stationary time series {Zt }.
(i) Define the autocovariance generating function of Zt . [2]
(ii) Suppose 
 1 k=0
1
γ(k) = k = ±1
 2
0 otherwise
Find the autocovariance generating function of Zt . [6]

page 2 of 3
 HSTS203

B7. (a) Suppose {Zt } is an ARMA(1,1) model expressed as

Zt = φZt−1 + at − θat−1

where {at } is a white noise process. Show that the autocorrelation function of Zt
is given by
(1 − θφ)(φ − θ) k−1
ρk = φ for k ≥ 1 [10]
1 − 2θφ + θ2
(b) Consider the second order moving average process {Zt } given by:

Zt = at + 0.7at−1 − 0.2at−2

where {at } is a white noise process. Find

(i) the mean of Zt . [1]
(ii) the variance of Zt . [2]
(iii) the autocorrelation function of Zt . [7]
(c) Consider the following AR(2) process
2 1
Zt = Zt−1 − Zt−2 + at
5 25
(i) Show that the model is stationary. [2]
(ii) Showing all your working, deduce that the autocorrelation function of Zt is
given by:
     k
24 1
ρk = 1 + k for k = 0, ±1, ±2, . . . [8]
26 5

B8. (a) For an AR(p) process given by:

Zt = φ1 Zt−1 + φ2 Zt−2 + . . . + φp Zt−p + at

(i) Determine the Yule-Walker equations for this process. [4]

(ii) Show how the method of moments is used to estimate the parameters φ1 to
φp . [8]
(b) Suppose a series Zt , Zt−1 , . . . from an AR(1) process given by

(1 − φ1 B)(Zt − µ) = at

(i) Find the l-step ahead forecast Ẑt (l) of Zt+l and the variance V ar[et (l)],
l = 1, 2, 3 where et (l) = Zt+l − Ẑt (l) is the l-step ahead forecast error. [10]
(ii) Show that Ẑt (l) is a minimum square error forecast. [8]

END OF EXAMINATION

page 3 of 3