n step rolling forecast using garch

[Nau2306]

Hi

I am using eviews 6. Is it possible to produce n-step rolling forecasts of the garch model in eviews? Is there a specific fucntion in eviews to carry it out?

Thanks

[EViews Gareth]

There is nothing built in, no.

[Nau2306]

Hmm is there any program available to do it in eviews? I have been trying to work on the codes i found on the forum itself but to no avail!!

[Nau2306]

Please find attached the workfile and program i have been trying. It is giving me 2 error messages:

1. error in sample : attempt to set sample outside of workfile range
structure/resize (from \workfile\proc menu) can expand the workfile range in 'SMPL 9/05/2016 9/06/2016'
My comments or queries: I have been estimating the data from 01/01/2000 to 05/03/2008 which represents the window= 3046. How come it is giving result for 2016??

2. WINSE is an illegal or reseerved name in 'DO_EQ1.FIT WIN8F [WINSE WINVAR]'
My comments or queries: I have tried to change the standard errors and variance series name but i am still getting the same error message.

Hope someone can help me.

Thanks

[EViews Gareth]

1. The sample error is from the line:

Code: Select all

smpl {%1pers} {%1pere}

It isn't particularly easy to follow your code, but I believe you have set %1pere to be the date associated with the value 1 + !i + !window. Since !i runs from 1 to !length-!window+1-!step, the final run of the iteration loop, will be:

Code: Select all

%1pere = date associated with 1+!length-!window+1-!step + !window = 1+!length

Since !length is the size of the workfile, you've just set %!pere to be the date associated with the first value outside of the workfile. That won't work.

2. You don't need the square brackets:

Code: Select all

EQ1.FIT WIN8F WINSE WINVAR

[Nau2306]

Well I have tried to produce 1 step ahead rolling forecast with the garch using as guide the code posted by eviews esther on may 25, 2011 where 4 period ahead forecast was generated for rolling regression. This is what was done to generate the 4 period ahead forecast.

' 4-period-ahead forecast
%4pers = @otod(@dtoo(%start)+!i+!window-1) 'start point
%4pere = @otod(@dtoo(%start)+!i+!window+2) 'end point: %4pere - %4pers +1 = 4
if {%end} < {%4pere} then 'check whether the forecast end point is greater than the workfile end point
return
endif

So i tried to adjust it to generate 1 period ahead forecast but dont know if its correct.

I am confused about one thing. With the eq1.fit function, am I forecasting the error or the variance??

Thanks

[EViews Gareth]

The fit function, as it states in the documentation, lets you forecast the mean equation's dependent variable, the standard error of that variable, and the GARCH conditional variance.

[Nau2306]

Ok thanks

Think I will have to work out the n-step ahead forecast and try to understand better how to generate it.

Thanks for your help

[Nau2306]

hi

I have tried a new code for the 1 step ahead garch (1,1) forecast but it is giving me the following error

error in sample: attempt to set sample outside of workfile range
Structure/Resize (from Workfile/Proc menu) can expand the workfile range in 'SMPL 12 12+3046-1

I know its got to do with the sample estimation but I cant see what's wrong. Could someone please look into the matter? i am attaching the program as well as the workfile.

Thanks

[EViews Gareth]

I just ran that program on the workfile, and it didn't error.

[Nau2306]

Oh did u run it on eviews6?

[EViews Gareth]

Yep.