IRF of a nonstationary system

Dear all:
In my model investment ,output is growing with technology ,and technology follows:log(A(t))=log(A(t-1))+mu_a+e_a(t) ,and another shock e_q in log(q(t))=q(t)=(1-rho_q)log(mu_q)+rho_qlog(q(t-1))+e_q(t)
the system is writen in difference form with dy=day/y(-1) and di=dai/i(-1)

now I have two questions:
1.if I simulate the model and want to get the irf of Y(t)=A(t)*y(t) ,I(t)=A(t)*i(t) do I only need to acumulated sum up the irf of di_e_a di_e_q dy_e_a dy_e_q in workspace(a series of number) ?

2.If I estimate the model and want to get posterior irf of Y(t)=A(t)*y(t) ,I(t)=A(t)*i(t) what should I do ?

I find baysian irf of variable in oo_.PosteriorIRF.dsge. but I don’t know how to read it.can anyone give me some hint ?what is response number and what is confidence interval number?

what is meaning block val(:,:,1,1),I have 10 endo vars and there is 10 column in each bolck

thank you all

  1. Yes, you just have to accumulate them
  2. This is tricky and I don’t know if this can be done easily. As IRFs are nonlinear functions of the parameters, the mean of the nonstationary IRFs is not equal to the sum of the mean IRFs for the stationary IFS plus the accumulated trend. Thus, you would have to hack Dynare to store the accumulated IRF for each parameter combination and compute the mean/median and the HPDIs yourself.

oo_.PosteriorIRF.dsge stores what the field-names say, namely the mean and median IRFs as well as the lower and upper bounds of the HPDI (“Bayesian confidence intervals”). What do you mean with