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

Add check for positive definiteness of Sigma_u^star 

Otherwise, complex likelihoods may occur
time-shift
Johannes Pfeifer 2017-04-09 12:43:10 +02:00 committed by Stéphane Adjemian (Charybdis)
parent d0682bca30
commit 52a082a1f9
2 changed files with 12 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

9
matlab/dsge_var_likelihood.m
View File

@ -233,6 +233,15 @@ assignin('base','GYX',GYX);
iGXX = inv(GXX);
PHI_star = iGXX*transpose(GYX); %formula (22), DS (2004)
SIGMA_u_star=GYY - GYX*PHI_star; %formula (23), DS (2004)
[SIGMA_u_star_is_positive_definite, penalty] = ispd(SIGMA_u_star);
if ~SIGMA_u_star_is_positive_definite
fval = Inf;
info(1) = 53;
info(4) = penalty;
exit_flag = 0;
return;
end
if ~isinf(dsge_prior_weight)% Evaluation of the likelihood of the dsge-var model when the dsge prior weight is finite.
tmp0 = dsge_prior_weight*NumberOfObservations*TheoreticalAutoCovarianceOfTheObservedVariables(:,:,1) + mYY ; %first term of square bracket in formula (29), DS (2004)
tmp1 = dsge_prior_weight*NumberOfObservations*GYX + mYX; %first element of second term of square bracket in formula (29), DS (2004)

4
matlab/print_info.m
View File

@ -122,7 +122,9 @@ if ~noprint
fprintf('\n The dsge_prior_weight is dsge_var=%5.4f, but must be at least %5.4f for the prior to be proper.\n',info(2),info(3));
error('You are estimating a DSGE-VAR model, but the value of the dsge prior weight is too low!')
case 52 %dsge_var_likelihood
error('You are estimating a DSGE-VAR model, but the implied covariance matrix of the VAR''s innovations is not positive definite!');
error('You are estimating a DSGE-VAR model, but the implied covariance matrix of the VAR''s innovations, based on artificial and actual sample is not positive definite!');
case 53 %dsge_var_likelihood
error('You are estimating a DSGE-VAR model, but the implied covariance matrix of the VAR''s innovations, based on the artificial sample, is not positive definite!');
case 55
error('Fast Kalman filter only works with stationary models [lik_init=1] or stationary observables for non-stationary models [lik_init=3]')
case 61 %Discretionary policy