Error Bands for IRF with Short and Long-Run Restrictions

[TomDoan]

This is an example of calculation of error bands for a structural VAR identified by short- and long-run restrictions. This is an example from the 2nd edition of the VAR e-course which describes its workings in detail. This is considerably simpler that the Bjornland-Leitemo example, and so is a better choice as the base program for adapting to a different model. It's an example out of Martin, Hurn and Harris, based upon a paper by Peersman.

Note that this requires RATS v9.

var_6_3.rpf

Program file

(2.41 KiB) Downloaded 1136 times

peersman_data.dat

Data file

(9.73 KiB) Downloaded 1082 times

[sebaschoo]

Hi Tom,
I'm using your program in order to estimate a structural VAR with seven variables but I would like to modify the magnitude of the shocks (I mean: setting for example a 2 standard deviation shock). Can you help with this issue?

Thank you!

[TomDoan]

I'm not sure what the point of that is. The IRF's for double the shock are double the IRF's for a single shock. You can adjust the FLongShort procedure to multiply the factor by two, but I would suggest not doing that.

[pfb382]

Hi Tom,

In the above given code to peersman: What does this line mean:

Code: Select all

estimate(sigma) 1979:5 *

and why the 1979:5? What does this specific observation do?

[TomDoan]

That means 1979:5 to the last possible entry (given the data).

This is an example out of Martin, Hurn and Harris. That's what they did and I think they copied that from Peersman's original article. If it's important, you would have to check that.