queries regarding certain time series functions

[aditi]

hi

I had a few queries regarding certain time series procedures and functions in EViews version 6. I would be greatly obliged if you could look into these queries.
1. Does EViews implement the 'Exact Maximum Likelihood' (EML) estimation technique under the Kalman filter estimation or under the general estimation techniques? Does a program sub-routine exist for the same or can it be performed using the logL object?

2.Is it possible to program the Beveridge nelson decomposition in EViews as it does not have a ready-to-use procedure for the same?

3. Can the 'Johansen normalization technique' used in cointegration be performed in EViews or does it have to be programed?

4. Can Spectral analysis involving finding the spectral windows, lag windows, etc. be done in EViews ?

Thanking you in anticipation,
Aditi

[EViews Glenn]

1. If the initial conditions satisfy the stability conditions and the relevant system matrices are non-time varying, we solve for them using the long-run relationship. In this case, the likelihood is exact. If these conditions are not met, then we use diffuse priors so the likelihood is approximate.

2. The BN decomposition can be programmed in EViews with a bit of work. EViews sspace objects will readily output the filtered series. Unfortunately, we don't have a single command way of getting the system matrices into matrix objects, but this can be done in a few steps. Once you have those two sets of output, you can solve for the BN decomposition.

3. EViews estimates Johansen cointegration methods.

4. We do not have good tools for spectral analysis.

[aditi]

hi

Thank you for replying.

A clarification about the first point: I need to estimate ARMA(p,q) (p=1,2,3, q=1,2,3) models using the exact ML estimation technique. How can it be estimated (can i program it using the logL object which would be extremely complicated for any orders beyond p=q=1)? Do you have access to a code for such an estimation?

Thanking you in anticipation,
Aditi