EViews.com

Dear all,
I have a question regarding how exactly the sspace handle missing data.
Based on my experience with other software, I can use a binary indicator (0 or 1) to indicate which observations are supposed to be omitted from the maximum likelihood estimation process.
Then, the ML process is modified to have two calculation processes, one without omitted observations and the other with omitted observations.
I wonder exactly how I can make the ML process in the sspace module to do the same way.

Thanks,
Tom99163

Unfortunately, you cannot edit the ML function in state space object. And I don't think you need to. State space model does a prediction for the missing value, but skips the correction step. Therefore, prediction variance is higher around missing values. You really do not have to omit missing values when modeling with the state space, unless there is a theoretical or empirical reason to do so.

Thanks, Trubador.

My state space model includes both monthly and quarterly series, which require the mixed frequencies setup.
I learnt from other software to treat the missing months in the quarterly series as missing observations.
Is there a way in the sspace module to handle the mixed frequencies of monthly and quarterly series?

Best,
Tom99163

Yes, that is correct. You can work in the monthly frequency and treat the missing values of quarterly series as state variables. As long as you define the measurement equation with respect to these state variables, Kalman filter will handle the rest. I do not know the details of your model, but it seems you have an opportunity to try MIDAS method, which is recently joined to EViews' toolbox.