SVAR crashes in 32 bits version

[Gerard]

Issue with EViews 9: On a 32 bits machine an SVAR with user defined A and B matrixes crashes. Using a Cholesky decomposition there is no problem.

On my 64 bits machine using the same model EViews 9 does not crash, but reports a problem with the Hessian matrix as it should.

[EViews Gareth]

Could you provide the workfile (and VAR specification)?

[Gerard]

The workfile is in your pm

[EViews Gareth]

Will be fixed next patch.

[Helia]

Hello!

I'm having exactly the same problem.When i try to introduce short run matrices (pata, patb),or my long run matrix (patc), my eviews 8 (student version) just crashes.
However, i can open and read eviwes workfiles with svar restrictions.
Please, find attached my file.

Thanks in advance,

Helia

[EViews Glenn]

What is the date on your EViews 8? (Help/About EViews)

[Helia]

What is the date on your EViews 8? (Help/About EViews)

I don't know if this is the right answer:
Student Version : June 24 2014
Serial Number: E8008719
Expires on : 2016-12-19

[EViews Jason]

Helia,

We cannot replicate the crash. We do get errors about the 'structural VAR matrix B is singular at starting values...' but EViews does not crash with the workfile you provided.

Please list the steps used to get the crash.

[Helia]

Yes, you are right! I forgot to mention one step. It is actually when i introduce the SR restriction i want to estimate (not Chololeski one as i sent in text ) that it crashes,and i cannot save it. To be more specific, in SVAR options, Identifying restrictions replace the last line "@e6 = C(16)*@e1 + C(17)*@e2 + C(18)*@e3 + C(19)*@e4 + C(20)*@e5 + C(21)*@u6" by "@e6 = C(21)*@u6". Once you press ok, it just stops running.That's why i tried with matrices instead. But it didn't work out neither.

And by the way, how can i reset the starting values in "Optimation control" in order to correct the problem with the Hessian of structural VAR? Could you help me?
From the last file i sent, i change the matrix B (patb), instead of a identity diagonal (nr 1 in diagional) i replace by "NA".

[EViews Jason]

The crash will be fixed in the next patch.

Not sure on what you are asking in the second part. Can you elaborate?

[Helia]

Thanks Jason,

I solve it, it was my mistake.

Just wondering, the SVAR add-in that help to estimate both short and long run restrictions, is it possible to have the variance decomposition?
And the output graphs, the dashed lines indicate a 95% confidence interval, or the one standard error bounds?

[EViews Glenn]

You'll have to ask the author of that add-in.