Am 20.02.2015 um 09:32 schrieb Paolo Chirico:
Dear Gretl users,
I discovered a way to perform (in Gretl) a local level model with
conditional heteroscedastic disturbances (see attached script).
The result is very similar to an ARIMA(0,1,1)+GARCH(1,1).
This way can be the base for several state space models with conditional
heteroscedastic disturbances.
Unfortunately, at the moment, I haven't a smart idea about the MLE
estimation of the parameters.
Any suggestions will be appreciated.
Hi Paolo,
I don't quite understand what you mean with that because you are using
the Kalman filter, and isn't that exactly there for getting the MLE?
Unrelated remark for Allin about syntax highlighting: Perhaps the
keywords like 'obsy' etc. as mentioned in 30.5 of the user guide should
be syntax-colored, like 'params' in mle blocks, 'weights' and
'orthog'
for GMM.
(And BTW, 'equations' in the plural form since gretl version 1.9.7 is
also still not colored.)
cheers,
sven