Why the Sspace model estimation is so volatile?

For technical questions regarding estimation of single equations, systems, VARs, Factor analysis and State Space Models in EViews. General econometric questions and advice should go in the Econometric Discussions forum.

Moderators: EViews Gareth, EViews Moderator

Alfred123
Posts: 31
Joined: Fri Jul 19, 2019 8:00 pm

Why the Sspace model estimation is so volatile?

Postby Alfred123 » Fri Aug 20, 2021 1:09 am

Dear all,

I constructed a state-space model to estimate the state variable. The model program works quite well at first. Picture 1 is the outcome I get from the estimation, the last observation of the model is June of this year. However, after I update the data to the July of this year, the outcome changed so dramatically, just as I showed in picture 2.

My model is a mix-frequency state-space model, which consists of four monthly variables and one quarterly variable. All of these variables are seasonal-adjusted then changed into their growth rates. I compare the seasonally adjusted data for every variable with and without the new observation. None of them show big differences. Therefore, it may not be the problem of the new observation or the seasonal adjustment.

So, I'd like to know why, and how to fix it. Thanks!
Attachments
pic_2.jpg
Picture 2, the outcome I think which is wrong.
pic_2.jpg (35.76 KiB) Viewed 9681 times
pic_1.jpg
Picture 1, the outcome I wanted.
pic_1.jpg (37.47 KiB) Viewed 9681 times

EViews Glenn
EViews Developer
Posts: 2671
Joined: Wed Oct 15, 2008 9:17 am

Re: Why the Sspace model estimation is so volatile?

Postby EViews Glenn » Tue Aug 24, 2021 9:55 pm

It is impossible to answer this question with the information provided. A workfile with estimated objects would help.

Alfred123
Posts: 31
Joined: Fri Jul 19, 2019 8:00 pm

Re: Why the Sspace model estimation is so volatile?

Postby Alfred123 » Tue Aug 24, 2021 11:40 pm

EViews Glenn wrote:It is impossible to answer this question with the information provided. A workfile with estimated objects would help.

Hi Glenn,
Thanks for the reply.
I post my workfile here.

There're two sheets in the workfile, one of them contains four monthly data and the other contains one quarterly data. The four monthly data are named "elec", "pt", "rail" and "freight" respectively. The quarterly data is named "emp". The suffix "X_a" means the accumulative value of the variable X; the suffix "X_int" means the interpolation value of the variable X; the suffix "X_sa" means the seasonal-adjusted value of the variable X. At last, "Xg" means the y-o-y growth rate of the variable X. The state space model is constructed with variables "freightg", "railg", "ptg", "elecg" and "empg".

The model estimation outputs are listed in the monthly data sheet.
Attachments
state_space_model.wf1
(748.61 KiB) Downloaded 127 times

EViews Glenn
EViews Developer
Posts: 2671
Joined: Wed Oct 15, 2008 9:17 am

Re: Why the Sspace model estimation is so volatile?

Postby EViews Glenn » Fri Aug 27, 2021 5:45 pm

The estimated sspace object is singular at the final values. I didn't delve too much into the specification, I don't think that your EMPG equation temporal disaggregation is working well.


Return to “Estimation”

Who is online

Users browsing this forum: Cscottmcwall-ea, Google [Bot] and 29 guests