284 News and risk
Table 12.6.Company news announcements and intraday conditional volatility of the S&P/ASX 200 Index and the SPI 200 Futures (multivariateGARCH results). The results are based on estimation of the following multivariate GARCH(1,1) model:Diagonal VECHMean equations:r^1¼tr^1tþ 1þ 1"^1tandr^2¼tr^2tþ 1þ 2"^2t;Variance equations:
211¼t!^11þ11(^2) " 1
t
1
þ
11
211
t
1
þ
^11
N
t
and
222
¼t
!^22
þ
22
(^2) " 2
t
1
þ
22
222
t
1
þ
^22
N
;t
Covariance equation:
212
¼t
!
12
þ
12
"^1
t
" 1
2 t
1
þ
12
212
t
1
þ
^12
N
;t
Diagonal BEKKMean equations:
r^1
¼t
r^1
t
þ 1
þ 1
"^1
t
and
r^2
¼t
r^2
t
þ 1
þ 2
"^2
t;
Variance equations:
211
¼t
!^11
þ
211
(^2) " 1
t
þ 1
211
(^2) 11
t
1
þ
^11
N
t
and
(^2) 22
¼t
!
22
þ
222
(^2) " 2
t
þ 1
222
222
t
þ 1
^22
N
;t
Covariance equation:
212
¼t
!
12
þ
11
22
"^1
t
" 1
2 t
^1
þ
11
22
(^2) 12
t
1
þ
^12
N
;t
in which
r^1
t
and
r^2
t
are the seasonally adjusted returns of the S&P/ASX 200 Index and the SPI 200 Futures at the
tth interval,
(^2) 1
t
and
(^2) 2
t
are
conditional variances of the error processes
"^1
andt
"^2
, respectively, andt
N
is the number of all company announcements in intervalt
t.
P
-values are
given in parentheses. ‘‘Wald Test’’ refers to the chi-square test statistics for the null hypothesis of
11
¼
^12
¼
^22
¼
- The results are presented for
the whole sample period (October 1, 2003 to September 30, 2009) and for the two subsample periods: October 1, 2003 to November 1, 2007 andNovember 2, 2007 to September 30, 2009.October 1, 2003 to September 30, 2009October 1, 2003 to November 1, 2007November 2, 2007 toSeptember 30, 2009^11^12^22Wald^11^12^22Wald^11
12^22WaldTestTestTestDiagonal VECH 3.010^33.110^33.310^3701.89 1.810^31.810^32.010^3926.14 0.013 0.012 0.012 454.72(0.00)(0.00)(0.00)(0.00)(0.00)(0.00)(0.00) (0.00) (0.00) (0.00) (0.00) (0.00)Diagonal BEKK 1.110^57.510^64.910^620.81 1.810^31.910^32.110^3854.14 0.013 0.013 0.013 407.34(0.00)(0.00)(0.00)(0.00)(0.00)(0.00)(0.00) (0.00) (0.00) (0.00) (0.00) (0.00)