Information Technology Reference
In-Depth Information
Residual-based tests are formulated to test the null hypothesis that the multiple time
series Y
are not cointegrated using the scalar unit root tests, such as the ADF test,
which are applied to the residuals
(
t
)
in (18.35)
In [26], the ADF test as well as two additional tests Z α and Z t , developed earlier
by Phillips [25], were applied to check for the presence of a unit root in the residuals
ξ t . In order to perform the unit root test, we fit an AR(1) model to
ξ t , t
=
1
,
2
,...
ξ t , t
=
1
,
2
,...
according to
ξ t = αξ t 1 + ρ t ,
t
=
1
,
2
,...
.
(18.36)
Then the statistic Z α in Phillips-Ouliaris test is defined as follows:
s Tl
s 2
1
2 ·
ρ
( α
Z α =
T
1
)
1 ,
(18.37)
1
T 2
t
2
t
ξ
=
2
whereas the Z t statistic is given by the following formula:
T
t = 2 ξ
1
2
s Tl
s 2
ρ
· ( α
)
s Tl
1
1
2 ·
2
Z t =
2 ,
(18.38)
s Tl T 2
1
t
1
t
2
ξ
=
2
t
where
T
t = 1 ρ
1
T
s 2
2
t
ρ =
,
(18.39)
T
t = 1 ρ
T
s = 1 w sl
T
1
T
2
T
s Tl =
2
t
+
t = s + 1 ρ t ρ t s ,
(18.40)
s
w sl =
1
1 .
(18.41)
l
+
Note that s 2
ρ
2
ρ
and s Tl are consistent estimators for the variance
σ
of
ρ t and the partial
lim T E T S T , where S T =
2
t
sum variance
σ
=
ξ t is the partial sum of the
=
1
error terms in (18.36).
The critical values for Z α and Z t statistics can be found in [26] (Tables I and II).
Phillips and Ouliaris tabulated the values for cointegrating regressions with at most 5
explanatory variables. Some estimates of the critical values for the Phillips-Ouliaris
test ( Z α ) are listed in Table 18.7.
Table 18.7: Critical values of the asymptotic distributions of the Z α statistic for test-
ing the null of no cointegration (Phillips-Ouliaris demeaned, reproduced from [26]).
Parameter n
(
n
=
K
1
)
represents the number of explanatory variables
n
90%
95%
99%
1
17 . 0390
20 . 4935
28 . 3218
2
22 . 1948
26 . 0943
34 . 1686
3
27 . 5846
32 . 0615
41 . 1348
4
32 . 7382
37 . 1508
47 . 5118
5
37 . 0074
41 . 9388
52 . 1723
 
Search WWH ::




Custom Search