The RATS Software Forum

Dear tom,

I hope this finds you well,

I try to apply the procedure @MCProcessIRF, to get the print out of the IRFs after a Gibbs sampling BVAR.

@MCProcessIRF(model=SBVAR_EG,center=median,percentile=||.16,.84||,IRF=ImpEgy, Lower= LowEgy, Upper=UpEgy)
print / ImpEgy

The results that I get are different than the plotted IRFs with @MCGraphIRF; when I plot ImpEgy I get different graphs!!

Would you give me some help on this.

Thank you very much

Works fine for me. Did you also use CENTER=MEDIAN on the @MCGRAPHIRF?

Yes, is this the reason?

@MCGraphIRF(model=SBVAR_EG,center=median,percentile=||.16,.84||,page=one),

Another question: if I use the option page = one, how could I have the titles of the graphs as the impact of a "shock" to a "Model variable"

Thank you

The graphs should be the same, just possibly with different scales.

You may be the first person ever to use PAGE=ONE. I don't recommend it since it doesn't standardize the scales. However, I added the headers to it and posted a new version at http://www.estima.com/forum/viewtopic.php?f=7&t=995.

Hi Tom,

A quick question. After estimating a VAR, I apply MCProcessIRF to get the IRF and Upper and Lower bands for each variable in the VAR.

@mcprocessirf(model=varmodel, percentiles=||.16, .84||, center=mean, lower=lower, upper=upper, irf=irf)

The IRFs produced by mcprocessirf...are they one-unit or standard deviation shocks?

Thanks,
Tom V.

TVolscho-286 wrote:Hi Tom,

A quick question. After estimating a VAR, I apply MCProcessIRF to get the IRF and Upper and Lower bands for each variable in the VAR.

@mcprocessirf(model=varmodel, percentiles=||.16, .84||, center=mean, lower=lower, upper=upper, irf=irf)

The IRFs produced by mcprocessirf...are they one-unit or standard deviation shocks?

Thanks,
Tom V.

That depends upon how you generated the %%responses. Unless you did something to do unit shocks, they will be standard deviation shocks.

Thanks Tom.

So from Montevar.rpf, if I wanted one-unit shocks, I would alter the factor option as below, correct?

impulse(noprint,model=model,factor=fsigmad,$
results=impulses,steps=steps)

into:

impulse(noprint,model=model,factor=%identity(nvar),$
results=impulses,steps=steps)

Yes.