and after reading some paper and code, then try, I find it is easily to show "missing value in the first obsevation"
I think it is because the first observation has not former perid value to calculate this perid's var and res, is it right?
and after some reading I think there is 2 way to deal with it...
like
1 :
' set sample (1/2/82-7/9/92) as in Table 1, column 1
' 10/6/86 is obs 3945
sample s0 2750 2750
sample s1 2751 5392
smpl s1
' set presample values of expressions in logl
smpl s0
series sig2 = omega(1)
series resma = 0
(that is give value to the first obsevation, is it right?)
2
use the dummy value
series d1=0
smpl
d1=1
smpl 1 1548
....
var=@recode(d1=1,omega(1)/1-alpha(1)-beta(1)),hu(1)*z0*var(-1)+hu(2)*z1*var(-1)+hu(3)*z2*var(-1)+du(1)*z0*z1*(res(-1)^2)+du(2)*z1*(res(-1)^2)+du(3)*z2*(res(-1)^2)+omega(1)+alpha(1)*(res(-1)^2+beta(1)*var(-1))
(Is it using the dummy value?)
so everyone, is my understanding right?
I know so little and hope you do not laugh at me...
Thank you very much.
and the programme is really not girl's job...I am exhausted about the stgarch model and missing value and other error .....I have tried all my best to deal with it....but...
