Академический Документы
Профессиональный Документы
Культура Документы
STATIONARY AND
NONSTATIONARY
VARIABLES
By:
Assoc. Prof. Dr. Sallahuddin Hassan
SEEQ5133
Applied Econometrics
DEFINITION OF
STATIONARY
DEFINITION OF
Stationary and non-stationary variables
STATIONARY
must be distinguished.
SEEQ5133
SPURIOUS REGRESSION
SPURIOUS REGRESSION
SEEQ5133
SEEQ5133
Hypothesis:
H0 : 1
H1 : 1
Non stationary
Stationary
Reject Ho if C
SEEQ5133
SEEQ5133
AUGMENTED DF
AUGMENTED DF
i 1
Yt i t
Yt Yt 1 i Yt i t
i 1
Yt t Yt 1 i Yt i t
i 1
SEEQ5133
COINTEGRATION
COINTEGRATION
Engle and Granger (1987) defined
cointegration as a state that if two (or
more) series are linked to form an
equilibrium relationship spanning in
the long run.
Cointegration implies that two random
variables, let says Yt and Xt, share
similar stochastic trends in common.
SEEQ5133
MATHEMATICAL APPROACH
OF COINTEGRATION
Y , X
Consider a set of two variables
that
are integrated of orderY ,1X (i.e.
)
~ I 1
and suppose that there
is a vector
1 , 2
which gives aYlinear
combination of
,X
which
denoted by:
1Yt is
2 Xstationary,
t t ~ I 0
[1]
MATHEMATICAL
APPROACH OF
If Equation [1] is normalized, it
COINTEGRATION
Y X
becomes
[2]
2
Xt
1
Yt
is the long-run or
equilibrium value of .
SEEQ5133
Cointegrating Vector
Cointegration in a
Bivariate VAR
'
Yt , X t
Yt for
(nonstationary) VAR
. That
Yt is
11 12 Yt 1 1t
X t 21 22
X t 1 2t
The matrix 11 is
1 given
12 by
1
21
22 1
SEEQ5133
Cointegration in a
Bivariate VAR
If cointegrating vector:
' 11 1 12
The matrix
is given by
1
' 21 11 1 12
11 1
SEEQ5133
TESTING FOR
COINTEGRATION
Cointegration in Single equation
SEEQ5133
SEEQ5133
The Engle-Granger
Approach
Yt 1 2 X t t
The Engle-Granger
Approach
t t 1 i t i t
1%
5%
10%
No Lags
-4.07
-3.37
-3.30
Lags
-3.73
-3.17
-2.91
SEEQ5133