stepan
/
dynare
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Fix for theoretical autocovariances for purely forward looking models 

In this case, Ex will by empty, leading to non-conformable empty matrices. Solution: compute correction term without using non-existent states.
time-shift
Johannes Pfeifer 2013-07-24 23:38:37 +02:00
parent 3e2a75d33a
commit 18664da6fe
1 changed files with 10 additions and 5 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

15
matlab/th_autocovariances.m
View File

@ -126,11 +126,16 @@ if options_.order == 2 || options_.hp_filter == 0
aa = ghx(iky,:);
bb = ghu(iky,:);
if options_.order == 2 % mean correction for 2nd order
Ex = (dr.ghs2(ikx)+dr.ghxx(ikx,:)*vx(:)+dr.ghuu(ikx,:)*M_.Sigma_e(:))/2;
Ex = (eye(n0)-AS(ikx,:))\Ex;
Gamma_y{nar+3} = NaN*ones(nvar, 1);
Gamma_y{nar+3}(stationary_vars) = AS(iky,:)*Ex+(dr.ghs2(iky)+dr.ghxx(iky,:)*vx(:)+...
dr.ghuu(iky,:)*M_.Sigma_e(:))/2;
if ~isempty(ikx)
Ex = (dr.ghs2(ikx)+dr.ghxx(ikx,:)*vx(:)+dr.ghuu(ikx,:)*M_.Sigma_e(:))/2;
Ex = (eye(n0)-AS(ikx,:))\Ex;
Gamma_y{nar+3} = NaN*ones(nvar, 1);
Gamma_y{nar+3}(stationary_vars) = AS(iky,:)*Ex+(dr.ghs2(iky)+dr.ghxx(iky,:)*vx(:)+...
dr.ghuu(iky,:)*M_.Sigma_e(:))/2;
else %no static and no predetermined
Gamma_y{nar+3} = NaN*ones(nvar, 1);
Gamma_y{nar+3}(stationary_vars) = (dr.ghs2(iky)+ dr.ghuu(iky,:)*M_.Sigma_e(:))/2;
end
end
end
if options_.hp_filter == 0