Time varying SVAR

[dakila]

Ali,

Unfortunately, all variables are dependent (endogenous) variables for TVSVAR. There is no independent (exogenous) variables except constant. Instead of TVSVAR you should consider TAR model.

[ali_Economist]

Hi,

I am wondering what will be the size of V Matrix if I have 4 instead of three variables ? will it be the same as mentioned in the document or will be add extra row or column ? Please guide.

Regards

Ali

[ali_Economist]

Please reply to my previous message, I want to include 4 variables instead of 3. If needed how I can add into the V Matrix equation given in the document?

Regards

Ali

[dakila]

If you choose 4 variables then the size of V matrix would be 4x4.

[ali_Economist]

Thank you very much, what about At the lower triangular matrix, given that we have 4 variables and also a constant, would it be 5 x 5 or 4x4?

Regards and best wishes

Ali

[dakila]

4x4

[ali_Economist]

Hi, once again need your help, it is giving near singular matrix. I got 5 variables and the stock market in the independent variable I am focusing on. Please help.

Regards

Ali

[adrangi]

Hi. I've reviewed the postings on the TVSVAR add-in. Very helpful, thanks. two issues:

1. I could not find the example data.

2. I'm using EV 9.5, my data runs from 2001m02-2016m02. I chose 2004m02, 2009m02 and 2016m02 for my date selection vector, no commas. I am getting the error message" Matrix vector index out of range." I noticed you already explained this on the forum!! Any ideas? Thanks much. Best, Bahram

[dakila]

1. For example C:\Users\...\Documents\EViews Addins\TVSVAR

2. Could you post the data?

[ali_Economist]

Hi,
how to put more than one options in the command line? can you please help me with format?

Regards

Ali

[adrangi]

Hi. Thanks much. Please see my worksheet copy

workfilepage.WF1

(145.01 KiB) Downloaded 2979 times

. I'm using the pc version of the variables, i.e., %change. I copied the page of the workfile so that you can even see the results I got after the error message. Appreciate any insights. Thanks much. Best, B

[dakila]

how to put more than one options in the command line?

For example:

Code: Select all

tvsvar(comp=1, mb=500) 2 40 datevec interest @ inflation unemployment interest

[dakila]

I am getting the error message" Matrix vector index out of range."

if you put 40 for the training sample, then you can't select 2004m02. Because your data starts with Nov 2001, first 40 obserbations (until Feb 2005) and lag will be used for the training sample. So 2004m02 is out of range.
If you choose 24 for the training sample, it is possible to include 2004m02. However, this number is too short to run regress.

For example, the following code works for me:

Code: Select all

svector dtsel=@wsplit("2004m02 2009m02 2016m02") tvsvar 2 24 dtsel pcco @ pcco pciag pcibz

[adrangi]

Thanks much. Makes perfect sense. I'll try it now. I may forget about the first date. Best, Bahram

[ali_Economist]

Hi,

Thanks for helping me on the previous issue. Now i got a problem and it is giving message "Matrix dimensions too large" if I drop a variable it works but then i don,t want to drop any variables and include all 5. I am trying to run this code.

svector dtsel=@wsplit("08/01/2008 01/02/2009 12/23/2016")
tvsvar(comp=1, mb=1000, nb=200, horizon=48) 2 102 dtsel lnjapan @ lnvnindex lnset lnse180interp hangseng lnjapan

Please help, Thanks in anticipation.

Regards

Ali