The RATS Software Forum

Posted: **Mon Aug 04, 2014 7:55 am**

Dear all
I estimate a Panel VAR model, the panel consists in 20 countries and 42 observations for each country, the dataset is perfectly balanced. Additionally, I have three variables (X1, X2 and X3), two lags and two deterministic variables (a country fixed effect and a country-specific linear trend) in the VAR.
I then calculate the impulse and the total cumulated impulse functions. Could you please help me telling me how I can compute the accumulated IRF plus their confidence bands?
For the accumulated IRF, I guess this code is sufficient:

Code: Select all

do j = 1,nbvar
accumulate irfm(j) / cirfm(j)
end do j

(where irfm the IRF and nbvar the number of variables, say 3 here). But how can I recover the confidence bands for the accumulated IRF ?
I hope you got my point.

Thank you in advance,

Best,
Romain

Posted: **Mon Aug 04, 2014 9:57 am**

After:

impulse(noprint,model=pvarmodel1,decomp=swish,results=impulses) nbvar nstep

and before the ewise responses(...), if you want all the responses accumulated, you can do

do i=1,nvar
do j=1,nvar
acc impulses(i,j)
end do j
end do i

The responses that are then saved and graphed will then be the accumulated ones.

Posted: **Mon Aug 04, 2014 10:24 am**

OK, many thanks, Tom.

The RATS Software Forum

Confidence bands of accumulated IRF

Confidence bands of accumulated IRF

Re: Confidence bands of accumulated IRF

Re: Confidence bands of accumulated IRF