Forecasting with DSGE - From already estimated parameters

Stochastic simulations
1

Hi!

This is my first post. I’m new using Dynare/Matlab and I’m a little stuck.

I used already estimated parameters for a closed economy DSGE model (3 equations, following Fuhrer, 2009)), then I used already estimated parameters to work a two economies DSGE (similar to Kulish and Rees, 2011), where I was working with a world economy and a small open economy. I was followoing the dynare user guide and was very useful.

However, I’m trying to use the first model described above (closed economy) to forecast (20 steps ahead) the output gap. I used forecast option, however it returned me a mean forecast equal zero. I used different initial values, dropped steady option, but still it gives me only zeros. Even I created a new variable, say x4=x(+1), to see if I get the 4 steps ahead forecast.

Please, if anyone can spot what I’m doing wrong or if you have any advice, I will appreciate it!

Cheers, Manuel.

2

I would need to see the codes.

3

Hi Johannes,

Thanks a lot. I attach you my code.

Parameters calibration values are following to Bekaert et al, (2010).

Hope to hear from you soon. Cheers,DSGEfore.mod (1.3 KB)

4

You must use the histval-block.

%--------------------------------------------------------------------------
%%%%%%%%%%%%%%%%%     Structural DSGE - Closed Economy    %%%%%%%%%%%%%%%%%
%--------------------------------------------------------------------------


%--------------------------------------
% VARIABLES
%--------------------------------------

var pi x i;

varexo u v;

%--------------------------------------
% PARAMETERS
%--------------------------------------

parameters mu gamma delta lambda rho omega theta sigmau sigmav;

%--------------------------------------
% CALIBRATION
%--------------------------------------

mu=0.39;
gamma=0.06;
delta=0.58;
lambda=0.13;
rho=0.72;
omega=1.53;
theta=0.00;
sigmau=0.71;
sigmav=0.71;

%--------------------------------------
% MODEL
%--------------------------------------

model;

pi = mu*pi(-1)+(1-mu)*pi(+1)+gamma*x+u;

x = delta*x(-1)+(1-delta)*x(+1)-lambda*(i-pi(+1))+v;

i = rho*i(-1)+(1-rho)*(omega*pi+theta*x);

end;

initval;

pi = 0.00;

x = 0.00;

i = 0.00;

end;

%--------------------------------------
% COMPUTATION
%--------------------------------------

check;

shocks;

var u = sigmau*sigmau;
var v = sigmav*sigmav;

end;

stoch_simul(irf=20);

histval;
x(0) = 10.00;
end;

forecast(periods=15);
5

Thanks a lot! I appreciate it.