load shocks in a perfect foresight model

[Ke]

Hi Michael and everyone,

I have a similar question as maile. The model is a perfect foresight one with 4 exogenous driving forces. 3 of them have 35 observations, 1 has 34 observations. Other than that, no more shocks (up to infinite future).

I wrote

shocks(shocks_file=shocks1);
var s1;
periods 1;
values 0;
var s3;
periods 1;
values 1.1;
var s3;
periods 1;
values 0.9;
var s4;
periods 1;
values 1.2;
end;

but it returns "Input argument "dyn_data_01" is undefined."
shocks1.mat contains all the observations of the shocks (driving forces).

If I input shocks period-by-period, it works!

Also, I'm wondering what's the algorithm that dynare use when it solves perfect foresight models? Forward shooting or reverse shooting?

If it is forward shooting, should I worry about the accuracy of the solution, in the sense that as we move along the stable saddle path from the initial value towards the steady state, the dynamics of the system actually drive variables away from the saddle path. Overtime it maybe quite off the true solution since computation errors are accumulated. For a RBC type of model, will this cause a problem?

Thanks a lot!
Ke

[MichelJuillard]

this syntax should work with the snapshot of version 3

Best

Michel

[Ke]

hi Michel,

Thank you for the kind remind. I did forget to update those files.

cheers,
Ke