endval for exogenous shocks in deterministic simulation

[stsp]

Hi,

I want to get the transition path for initial TFP process (exogenous variable) with 0<rho<1:

Code: Select all

z = rho*z(-1);

when the process changes permanently in period 2 (and this is known in period 1) to:

Code: Select all

z = rho*z(-1) + sigma;

What i did is the following:

Code: Select all

model block: z = rho*z(-1) + sigma + eps*sigma; (the NEW process)
initval: z = 0;
endval: z=(1-rho)*sigma;

shocks in eps:
periods 1 2:1000;
values -1 0; %the -1 is to keep z(1)=0;

When I get the results I see that z(2)=sigma/(1-rho); What did i do wrong? I thought it would calculate: z(2)=rho*z(-1) + sigma = sigma; but id doesnt. I havent used steady either.

(I know that I could get the correct result by setting the initval: z=0 and then add a shock eps=1 for period 2:1000, but i want to understand how to model it with initval and endval)

Thanks for your help!

[jpfeifer]

For endogenous variables, initval and endval are for setting initial and terminal conditions. Everything else will be endogenously computed. As z is purely backward looking, the only things that matter are the initial value for z you assign and the subsequent shock values. What you are not telling is what you did with sigma.