The RATS Software Forum

Hi Tom,

I have a question.

So, I am using the long run monetary neutrality assumption as in BQ. I use the following code:
estimate
compute bqfactor=%bqfactor(%sigma,%varlagsums)
compute implabel = || "Real GDP","Interest Rate","Real Money","Nominal Money" ||
impulses(model=bqvar,result=impulses,factor=bqfactor,steps=ksteps)
@varirf(model=bqvar,footer="Responses to Cholesky factor shocks", steps=ksteps,varlabels=implabel, accum=||1, 2, 3, 4||)

Now the irf's I get here for real variables do not converge to zero in the long run, but if I dont accumulate them, then even though they are very fluctuating but they converge to zero in the long run. Alternatively I have also done this:

impulse(model=bqvar,result=irfs,factor=bqfactor, steps=ksteps)
dec rect[ser] accumimp(%nvar,%nvar)
do i=1,%nvar
do j=1,%nvar
accumulate irfs(i,j) 1 ksteps accumimp(i,j)
end do i
end do j

How do I draw the IRF's after accumulating them in this way? I want to see if now my real variables converge to zero in the long run or not.

Thank you so much

If the variables are (log) differenced, then the unaccumulated IRF's will converge to zero as they will be stationary processes. However, the B-Q restriction is that certain of the responses sum to zero---in particular, the responses of the first three variables to the fourth shock will do that. If you're not seeing that, you're not looking at enough steps.

The most recent version of @MCGRAPHIRF has an option for accumulating the shocks after the full set of draws.

Hi Tom,

I still have a query. So I have accumulated the shocks using the following

impulse(model=bqvar,result=impulses,factor=bqfactor, steps=ksteps)
dec rect[ser] accumimp(%nvar,%nvar)
do i=1,%nvar
do j=1,%nvar
accumulate impulses(i,j) 1 ksteps accumimp(i,j)
end do i
end do j


On the series window I do get accumimp(1,4) accumimp(2,4) and so on..
I get a tabular form of the impuslses or shocks
How do I get the graphs of IRF's? What's the command for drawing the graphs of IRF's in this case after accumulation. If I do

graph(nodates,number=0,footer=$
"Response of Real GDP to a shock in nominal money")
# accumimp(1,4)

Is this fine Tom?

Many Many Thanks

That looks correct.