Re: [Gretl-users] VAR model and forecast
On 12/28/2011 11:48 PM, alexkakashi(a)libero.it wrote:
Hi,
I have two questions about the parameters of the VAR models. Let us (X,Y) a
bivariate time series and consider a VAR(1). I use
system method=sur
equation X const X(-1) Y(-1)
equation Y const X(-1) Y(-1)
end system
I have two questions. The first: can I save the parameter of X(-1) of the
first equation? I should use it in a bootstrap procedure.
Check out the '$sysA' accessor (see the function reference). In this
case the coefficient will probably sit in $sysA[1,1], but you'll have to
verify that with a test case.
The second: I'd like forecast the variables X, so I have used
the
fcast --out-of-sample --static X
Can I save this forecasts in a new variable?
Should be possible by doing
series theforecast = $fcast
or something like that.
hth,
sven