some confusion about elasticities from cumulated irf

[Fregensburg]

Good morning,

I'm using a recursive var to look at the exchange rate pass through to inflation.
With reference to the workfile, I can get some accumulated impulse response functions, with the following lines:

smpl @all
smpl @first 1999:02
prezirp.impulse(36, a, imp = chol, se = a) d1ylncpi d1ylnwpi d1ylnimpp oilinflocalcur d1ylnneer @ d1ylnneer @ oilinflocalcur d1ylnip_sa d1ylnneer d1ylnimpp d1ylnwpi d1ylncpi

Sure, the ordering may be a little arbitrary, but it seems plausible to me (oil price inflation, industrial production, annualised percentage change in the neer, import price inflation, wholesale price inflation and cpi inflation), at least for a first go.
To keep things simple in the line above I only shock the variables with a 1st deviation change in the 12 months log difference of the nominal effective exchange rate.

What I want to do is get the pass-through elasticity. In the literature, people compute it like this:
PT(t,t+j) = P(t,t+j / E(t,t+j)
with P being the cumulative percentage change in the cpi and E the cumulative percentage change in the NEER from month t through t+j periods after the shock.

By the way, this is a case of me not being entirely sure of what I have to do (both in terms of econometrics and in terms of programming)
So Im not sure how people compute this (I presume that the cumulative percentage change in the cpi is simply the response of d1ylncpi to d1ylnneer), but then what about the cumulative percentage change of neer (the response of d1ylnneer to what shock exactly? its own?)
What do you guys think? any thoughts from the pros -and perhaps some shorthand on the code to calculate this would be highly appreciated.
Many thanks