# Theoretical moments do not match data

Hi,
I have estimated a DSGE model with ML using my own MATLAB code. I have then included in dynare the estimated parameters to check whether the theoretical second moments from the estimated DSGE model are close to the data. I find that the theoretical moments are far greater and cannot see why. Can anyone suggest what I am doing wrong?
Details are below.
Thanks. Vito

My mod code is

// NEW KEYNESIAN MODEL

var r y p uy up ur;

varexo ey ep er;

parameters gamma gamma1 gamma2 phi del del1 del2 kappa psip1 psiy1 psir1 psip2 psiy2 psir2;

gamma=0.3253;
gamma1=0.8491;
gamma2=-0.1906;
phi=0.0076;
del=0.4943;
del1=0.2290;
del2=0.2767;

kappa=0.0243;

psip1=0.00025;
psiy1=0.1152;
psir1=0.6095;
psip2=0.2048;
psiy2=0;
psir2=0.2537;

model(linear);
y = gammay(1)+gamma1y(-1)+gamma2y(-2)-phi(r-p(1))+uy;
p = delp(1)+del1p(-1)+del2p(-2)+kappay+up;
r = psip1p(-1)+ psiy1y(-1)+ psir1r(-1) +psip2p(-2) + psiy2y(-2)+ psir2r(-2) + ur;
up=1.12387427199994ep;
uy=-0.0382815889959732
ep+0.759252556400376ey;
ur=0.221657786793378
ep+0.403195872651417ey+1.28914940964670er;
end;

shocks;
var ey; stderr 1;
var ep; stderr 1;
var er; stderr 1;

end;

steady(solve_algo=1); // calculate steady states (solve_algo=0 use FSOLVE, solve_algo=1 us DYNARE SOLVER)

check; // check the eigenvalues of the system

stoch_simul(periods=0,irf=0,simul_seed=3) y r p;

This code returns the following theoretical moments
VARIABLE MEAN STD. DEV. VARIANCE
y 0.0000 4.7862 22.9079
r 0.0000 15.0979 227.9472
p 0.0000 10.4777 109.7812

while the sample variances for y, r and p are 5, 6 and 13 respectively .

This is a well-known problem, which has given rise to endogenous priors, see ideas.repec.org/a/eee/moneco/v55y2008i7p1191-1208.html or the recent JEDC paper of Lombardi and Nicoletti