2016-12-15 10:36:16 +01:00
|
|
|
function [alphahat,etahat,epsilonhat,ahat,SteadyState,trend_coeff,aK,T,R,P,PK,decomp,trend_addition,state_uncertainty,M_,oo_,options_,bayestopt_] = DsgeSmoother(xparam1,gend,Y,data_index,missing_value,M_,oo_,options_,bayestopt_,estim_params_)
|
2017-05-16 15:10:20 +02:00
|
|
|
% Estimation of the smoothed variables and innovations.
|
|
|
|
%
|
|
|
|
% INPUTS
|
|
|
|
% o xparam1 [double] (p*1) vector of (estimated) parameters.
|
2008-03-03 12:10:16 +01:00
|
|
|
% o gend [integer] scalar specifying the number of observations ==> varargin{1}.
|
2016-06-01 08:23:27 +02:00
|
|
|
% o data [double] (n*T) matrix of data.
|
2008-11-25 18:31:37 +01:00
|
|
|
% o data_index [cell] 1*smpl cell of column vectors of indices.
|
|
|
|
% o missing_value 1 if missing values, 0 otherwise
|
2016-12-15 10:36:16 +01:00
|
|
|
% o M_ [structure] decribing the model
|
|
|
|
% o oo_ [structure] storing the results
|
|
|
|
% o options_ [structure] describing the options
|
|
|
|
% o bayestopt_ [structure] describing the priors
|
|
|
|
% o estim_params_ [structure] characterizing parameters to be estimated
|
2017-05-16 15:10:20 +02:00
|
|
|
%
|
2016-04-10 19:18:41 +02:00
|
|
|
% OUTPUTS
|
|
|
|
% o alphahat [double] (m*T) matrix, smoothed endogenous variables (a_{t|T}) (decision-rule order)
|
|
|
|
% o etahat [double] (r*T) matrix, smoothed structural shocks (r>=n is the number of shocks).
|
2007-04-15 16:21:11 +02:00
|
|
|
% o epsilonhat [double] (n*T) matrix, smoothed measurement errors.
|
2016-04-10 19:18:41 +02:00
|
|
|
% o ahat [double] (m*T) matrix, updated (endogenous) variables (a_{t|t}) (decision-rule order)
|
|
|
|
% o SteadyState [double] (m*1) vector specifying the steady state level of each endogenous variable (declaration order)
|
2007-04-15 16:21:11 +02:00
|
|
|
% o trend_coeff [double] (n*1) vector, parameters specifying the slope of the trend associated to each observed variable.
|
2016-04-10 19:18:41 +02:00
|
|
|
% o aK [double] (K,n,T+K) array, k (k=1,...,K) steps ahead
|
|
|
|
% filtered (endogenous) variables (decision-rule order)
|
2008-03-03 12:10:16 +01:00
|
|
|
% o T and R [double] Matrices defining the state equation (T is the (m*m) transition matrix).
|
2016-04-10 19:18:41 +02:00
|
|
|
% o P: (m*m*(T+1)) 3D array of one-step ahead forecast error variance
|
|
|
|
% matrices (decision-rule order)
|
|
|
|
% o PK: (K*m*m*(T+K)) 4D array of k-step ahead forecast error variance
|
|
|
|
% matrices (meaningless for periods 1:d) (decision-rule order)
|
|
|
|
% o decomp (K*m*r*(T+K)) 4D array of shock decomposition of k-step ahead
|
|
|
|
% filtered variables (decision-rule order)
|
2017-05-16 15:10:20 +02:00
|
|
|
% o trend_addition [double] (n*T) pure trend component; stored in options_.varobs order
|
2016-10-31 17:59:28 +01:00
|
|
|
% o state_uncertainty [double] (K,K,T) array, storing the uncertainty
|
|
|
|
% about the smoothed state (decision-rule order)
|
2016-12-15 10:36:16 +01:00
|
|
|
% o M_ [structure] decribing the model
|
|
|
|
% o oo_ [structure] storing the results
|
|
|
|
% o options_ [structure] describing the options
|
|
|
|
% o bayestopt_ [structure] describing the priors
|
2017-05-16 15:10:20 +02:00
|
|
|
%
|
2016-04-10 19:18:41 +02:00
|
|
|
% Notes:
|
|
|
|
% m: number of endogenous variables (M_.endo_nbr)
|
|
|
|
% T: number of Time periods (options_.nobs)
|
|
|
|
% r: number of strucural shocks (M_.exo_nbr)
|
|
|
|
% n: number of observables (length(options_.varobs))
|
|
|
|
% K: maximum forecast horizon (max(options_.nk))
|
2017-05-16 15:10:20 +02:00
|
|
|
%
|
2016-04-10 19:18:41 +02:00
|
|
|
% To get variables that are stored in decision rule order in order of declaration
|
|
|
|
% as in M_.endo_names, ones needs code along the lines of:
|
|
|
|
% variables_declaration_order(dr.order_var,:) = alphahat
|
2017-05-16 15:10:20 +02:00
|
|
|
%
|
|
|
|
% Defines bayestopt_.mf = bayestopt_.smoother_mf (positions of observed variables
|
|
|
|
% and requested smoothed variables in decision rules (decision rule order)) and
|
2016-04-10 20:53:13 +02:00
|
|
|
% passes it back via global variable
|
2017-05-16 15:10:20 +02:00
|
|
|
%
|
|
|
|
% ALGORITHM
|
|
|
|
% Diffuse Kalman filter (Durbin and Koopman)
|
2005-02-18 20:54:39 +01:00
|
|
|
%
|
2007-04-15 16:21:11 +02:00
|
|
|
% SPECIAL REQUIREMENTS
|
2008-03-03 12:10:16 +01:00
|
|
|
% None
|
2008-08-01 14:40:33 +02:00
|
|
|
|
2020-01-10 17:55:57 +01:00
|
|
|
% Copyright (C) 2006-2020 Dynare Team
|
2008-08-01 14:40:33 +02:00
|
|
|
%
|
|
|
|
% This file is part of Dynare.
|
|
|
|
%
|
|
|
|
% Dynare is free software: you can redistribute it and/or modify
|
|
|
|
% it under the terms of the GNU General Public License as published by
|
|
|
|
% the Free Software Foundation, either version 3 of the License, or
|
|
|
|
% (at your option) any later version.
|
|
|
|
%
|
|
|
|
% Dynare is distributed in the hope that it will be useful,
|
|
|
|
% but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
|
|
% MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
|
|
|
% GNU General Public License for more details.
|
|
|
|
%
|
|
|
|
% You should have received a copy of the GNU General Public License
|
|
|
|
% along with Dynare. If not, see <http://www.gnu.org/licenses/>.
|
2008-03-03 12:10:16 +01:00
|
|
|
|
2010-01-05 11:46:10 +01:00
|
|
|
alphahat = [];
|
|
|
|
etahat = [];
|
|
|
|
epsilonhat = [];
|
2009-12-16 18:17:34 +01:00
|
|
|
ahat = [];
|
|
|
|
SteadyState = [];
|
|
|
|
trend_coeff = [];
|
|
|
|
aK = [];
|
|
|
|
T = [];
|
|
|
|
R = [];
|
|
|
|
P = [];
|
|
|
|
PK = [];
|
|
|
|
decomp = [];
|
2014-07-15 11:13:41 +02:00
|
|
|
vobs = length(options_.varobs);
|
2009-12-16 18:17:34 +01:00
|
|
|
smpl = size(Y,2);
|
2005-02-18 20:54:39 +01:00
|
|
|
|
2015-06-09 14:51:32 +02:00
|
|
|
if ~isempty(xparam1) %not calibrated model
|
|
|
|
M_ = set_all_parameters(xparam1,estim_params_,M_);
|
|
|
|
end
|
2005-02-18 20:54:39 +01:00
|
|
|
|
2009-12-16 18:17:34 +01:00
|
|
|
%------------------------------------------------------------------------------
|
|
|
|
% 2. call model setup & reduction program
|
|
|
|
%------------------------------------------------------------------------------
|
2017-05-16 15:10:20 +02:00
|
|
|
oldoo.restrict_var_list = oo_.dr.restrict_var_list;
|
2015-04-03 17:59:51 +02:00
|
|
|
oldoo.restrict_columns = oo_.dr.restrict_columns;
|
|
|
|
oo_.dr.restrict_var_list = bayestopt_.smoother_var_list;
|
|
|
|
oo_.dr.restrict_columns = bayestopt_.smoother_restrict_columns;
|
|
|
|
|
|
|
|
[T,R,SteadyState,info,M_,options_,oo_] = dynare_resolve(M_,options_,oo_);
|
|
|
|
|
2017-08-10 15:47:51 +02:00
|
|
|
if info~=0
|
|
|
|
print_info(info,options_.noprint, options_);
|
|
|
|
return
|
|
|
|
end
|
2015-04-03 17:59:51 +02:00
|
|
|
oo_.dr.restrict_var_list = oldoo.restrict_var_list;
|
|
|
|
oo_.dr.restrict_columns = oldoo.restrict_columns;
|
|
|
|
|
2016-04-12 11:36:43 +02:00
|
|
|
%get location of observed variables and requested smoothed variables in
|
|
|
|
%decision rules
|
|
|
|
bayestopt_.mf = bayestopt_.smoother_var_list(bayestopt_.smoother_mf);
|
2009-12-16 18:17:34 +01:00
|
|
|
if options_.noconstant
|
2014-07-15 11:13:41 +02:00
|
|
|
constant = zeros(vobs,1);
|
2009-12-16 18:17:34 +01:00
|
|
|
else
|
2014-01-30 13:00:25 +01:00
|
|
|
if options_.loglinear
|
2009-12-16 18:17:34 +01:00
|
|
|
constant = log(SteadyState(bayestopt_.mfys));
|
|
|
|
else
|
|
|
|
constant = SteadyState(bayestopt_.mfys);
|
|
|
|
end
|
|
|
|
end
|
2014-07-15 11:13:41 +02:00
|
|
|
trend_coeff = zeros(vobs,1);
|
2009-12-16 18:17:34 +01:00
|
|
|
if bayestopt_.with_trend == 1
|
2015-03-01 15:12:34 +01:00
|
|
|
[trend_addition, trend_coeff] =compute_trend_coefficients(M_,options_,vobs,gend);
|
|
|
|
trend = constant*ones(1,gend)+trend_addition;
|
2009-12-16 18:17:34 +01:00
|
|
|
else
|
2015-03-01 15:40:47 +01:00
|
|
|
trend_addition=zeros(size(constant,1),gend);
|
2005-09-11 11:38:52 +02:00
|
|
|
trend = constant*ones(1,gend);
|
2009-12-16 18:17:34 +01:00
|
|
|
end
|
|
|
|
start = options_.presample+1;
|
|
|
|
np = size(T,1);
|
2016-04-12 11:36:43 +02:00
|
|
|
mf = bayestopt_.mf;
|
2009-12-16 18:17:34 +01:00
|
|
|
% ------------------------------------------------------------------------------
|
|
|
|
% 3. Initial condition of the Kalman filter
|
|
|
|
% ------------------------------------------------------------------------------
|
2017-05-16 15:10:20 +02:00
|
|
|
%
|
|
|
|
% Here, Pinf and Pstar are determined. If the model is stationary, determine
|
2016-04-10 19:18:41 +02:00
|
|
|
% Pstar as the solution of the Lyapunov equation and set Pinf=[] (Notation follows
|
|
|
|
% Koopman/Durbin (2003), Journal of Time Series Analysis 24(1))
|
2009-12-16 18:17:34 +01:00
|
|
|
%
|
|
|
|
Q = M_.Sigma_e;
|
|
|
|
H = M_.H;
|
|
|
|
|
2011-06-23 23:39:15 +02:00
|
|
|
if isequal(H,0)
|
2014-07-15 11:13:41 +02:00
|
|
|
H = zeros(vobs,vobs);
|
2011-06-23 23:39:15 +02:00
|
|
|
end
|
|
|
|
|
2016-06-16 16:25:11 +02:00
|
|
|
Z = zeros(vobs,size(T,2));
|
|
|
|
for i=1:vobs
|
|
|
|
Z(i,mf(i)) = 1;
|
|
|
|
end
|
|
|
|
|
|
|
|
expanded_state_vector_for_univariate_filter=0;
|
2009-12-16 18:17:34 +01:00
|
|
|
kalman_algo = options_.kalman_algo;
|
2010-01-05 11:46:10 +01:00
|
|
|
if options_.lik_init == 1 % Kalman filter
|
2009-12-16 18:17:34 +01:00
|
|
|
if kalman_algo ~= 2
|
|
|
|
kalman_algo = 1;
|
|
|
|
end
|
2016-12-18 18:23:42 +01:00
|
|
|
Pstar=lyapunov_solver(T,R,Q,options_);
|
2010-01-05 11:46:10 +01:00
|
|
|
Pinf = [];
|
2013-06-28 16:26:53 +02:00
|
|
|
elseif options_.lik_init == 2 % Old Diffuse Kalman filter
|
2009-12-16 18:17:34 +01:00
|
|
|
if kalman_algo ~= 2
|
|
|
|
kalman_algo = 1;
|
|
|
|
end
|
|
|
|
Pstar = options_.Harvey_scale_factor*eye(np);
|
2010-01-05 11:46:10 +01:00
|
|
|
Pinf = [];
|
2013-06-28 16:26:53 +02:00
|
|
|
elseif options_.lik_init == 3 % Diffuse Kalman filter
|
2009-12-16 18:17:34 +01:00
|
|
|
if kalman_algo ~= 4
|
|
|
|
kalman_algo = 3;
|
2016-06-16 16:25:11 +02:00
|
|
|
else
|
|
|
|
if ~all(all(abs(H-diag(diag(H)))<1e-14))% ie, the covariance matrix is diagonal...
|
2017-05-16 15:10:20 +02:00
|
|
|
%Augment state vector (follows Section 6.4.3 of DK (2012))
|
2016-06-16 16:25:11 +02:00
|
|
|
expanded_state_vector_for_univariate_filter=1;
|
|
|
|
T = blkdiag(T,zeros(vobs));
|
|
|
|
np = size(T,1);
|
|
|
|
Q = blkdiag(Q,H);
|
|
|
|
R = blkdiag(R,eye(vobs));
|
|
|
|
H = zeros(vobs,vobs);
|
|
|
|
Z = [Z, eye(vobs)];
|
|
|
|
end
|
2009-12-16 18:17:34 +01:00
|
|
|
end
|
2017-11-16 16:55:58 +01:00
|
|
|
[Pstar,Pinf] = compute_Pinf_Pstar(mf,T,R,Q,options_.qz_criterium);
|
2013-06-28 16:26:53 +02:00
|
|
|
elseif options_.lik_init == 4 % Start from the solution of the Riccati equation.
|
2020-01-10 17:55:57 +01:00
|
|
|
Pstar = kalman_steady_state(transpose(T),R*Q*transpose(R),transpose(build_selection_matrix(mf,np,vobs)),H);
|
2011-06-23 23:39:15 +02:00
|
|
|
Pinf = [];
|
|
|
|
if kalman_algo~=2
|
|
|
|
kalman_algo = 1;
|
|
|
|
end
|
2013-06-28 16:26:53 +02:00
|
|
|
elseif options_.lik_init == 5 % Old diffuse Kalman filter only for the non stationary variables
|
|
|
|
[eigenvect, eigenv] = eig(T);
|
|
|
|
eigenv = diag(eigenv);
|
|
|
|
nstable = length(find(abs(abs(eigenv)-1) > 1e-7));
|
|
|
|
unstable = find(abs(abs(eigenv)-1) < 1e-7);
|
|
|
|
V = eigenvect(:,unstable);
|
|
|
|
indx_unstable = find(sum(abs(V),2)>1e-5);
|
|
|
|
stable = find(sum(abs(V),2)<1e-5);
|
|
|
|
nunit = length(eigenv) - nstable;
|
|
|
|
Pstar = options_.Harvey_scale_factor*eye(np);
|
|
|
|
if kalman_algo ~= 2
|
|
|
|
kalman_algo = 1;
|
|
|
|
end
|
|
|
|
R_tmp = R(stable, :);
|
|
|
|
T_tmp = T(stable,stable);
|
2016-12-18 18:23:42 +01:00
|
|
|
Pstar_tmp=lyapunov_solver(T_tmp,R_tmp,Q,DynareOptions);
|
2013-06-28 16:26:53 +02:00
|
|
|
Pstar(stable, stable) = Pstar_tmp;
|
|
|
|
Pinf = [];
|
2009-12-16 18:17:34 +01:00
|
|
|
end
|
2010-03-23 18:17:15 +01:00
|
|
|
kalman_tol = options_.kalman_tol;
|
2015-04-03 17:48:25 +02:00
|
|
|
diffuse_kalman_tol = options_.diffuse_kalman_tol;
|
2010-03-23 18:17:15 +01:00
|
|
|
riccati_tol = options_.riccati_tol;
|
|
|
|
data1 = Y-trend;
|
2009-12-16 18:17:34 +01:00
|
|
|
% -----------------------------------------------------------------------------
|
|
|
|
% 4. Kalman smoother
|
|
|
|
% -----------------------------------------------------------------------------
|
2011-01-13 21:50:26 +01:00
|
|
|
|
|
|
|
if ~missing_value
|
|
|
|
for i=1:smpl
|
2014-07-15 11:13:41 +02:00
|
|
|
data_index{i}=(1:vobs)';
|
2011-01-13 21:50:26 +01:00
|
|
|
end
|
|
|
|
end
|
|
|
|
|
2015-10-13 17:26:39 +02:00
|
|
|
ST = T;
|
|
|
|
R1 = R;
|
2011-01-13 21:50:26 +01:00
|
|
|
|
|
|
|
if kalman_algo == 1 || kalman_algo == 3
|
2016-10-31 17:59:28 +01:00
|
|
|
[alphahat,epsilonhat,etahat,ahat,P,aK,PK,decomp,state_uncertainty] = missing_DiffuseKalmanSmootherH1_Z(ST, ...
|
2011-01-13 21:50:26 +01:00
|
|
|
Z,R1,Q,H,Pinf,Pstar, ...
|
2014-07-15 11:13:41 +02:00
|
|
|
data1,vobs,np,smpl,data_index, ...
|
2016-10-31 17:59:28 +01:00
|
|
|
options_.nk,kalman_tol,diffuse_kalman_tol,options_.filter_decomposition,options_.smoothed_state_uncertainty);
|
2011-11-02 14:02:12 +01:00
|
|
|
if isinf(alphahat)
|
2011-01-13 21:50:26 +01:00
|
|
|
if kalman_algo == 1
|
2020-07-15 11:00:48 +02:00
|
|
|
fprintf('\nDsgeSmoother: Switching to univariate filter. This may be a sign of stochastic singularity.\n')
|
2009-12-16 18:17:34 +01:00
|
|
|
kalman_algo = 2;
|
2011-01-13 21:50:26 +01:00
|
|
|
elseif kalman_algo == 3
|
2020-07-15 11:00:48 +02:00
|
|
|
fprintf('\nDsgeSmoother: Switching to univariate filter. This may be a sign of stochastic singularity.\n')
|
2011-01-13 21:50:26 +01:00
|
|
|
kalman_algo = 4;
|
2009-12-16 18:17:34 +01:00
|
|
|
else
|
2011-01-13 21:50:26 +01:00
|
|
|
error('This case shouldn''t happen')
|
2009-12-16 18:17:34 +01:00
|
|
|
end
|
2011-01-13 21:50:26 +01:00
|
|
|
end
|
|
|
|
end
|
2011-02-04 17:17:48 +01:00
|
|
|
|
2011-01-13 21:50:26 +01:00
|
|
|
if kalman_algo == 2 || kalman_algo == 4
|
2017-05-16 15:10:20 +02:00
|
|
|
if ~all(all(abs(H-diag(diag(H)))<1e-14))% ie, the covariance matrix is diagonal...
|
|
|
|
if ~expanded_state_vector_for_univariate_filter
|
|
|
|
%Augment state vector (follows Section 6.4.3 of DK (2012))
|
|
|
|
expanded_state_vector_for_univariate_filter=1;
|
|
|
|
Z = [Z, eye(vobs)];
|
|
|
|
ST = blkdiag(ST,zeros(vobs));
|
|
|
|
np = size(ST,1);
|
|
|
|
Q = blkdiag(Q,H);
|
|
|
|
R1 = blkdiag(R,eye(vobs));
|
|
|
|
if kalman_algo == 4
|
|
|
|
%recompute Schur state space transformation with
|
|
|
|
%expanded state space
|
|
|
|
[Pstar,Pinf] = compute_Pinf_Pstar(mf,ST,R1,Q,options_.qz_criterium);
|
2016-06-16 16:25:11 +02:00
|
|
|
else
|
2017-05-16 15:10:20 +02:00
|
|
|
Pstar = blkdiag(Pstar,H);
|
|
|
|
if ~isempty(Pinf)
|
|
|
|
Pinf = blkdiag(Pinf,zeros(vobs));
|
|
|
|
end
|
2016-06-16 16:25:11 +02:00
|
|
|
end
|
2017-05-16 15:10:20 +02:00
|
|
|
%now reset H to 0
|
|
|
|
H = zeros(vobs,vobs);
|
|
|
|
else
|
|
|
|
%do nothing, state vector was already expanded
|
2009-12-16 18:17:34 +01:00
|
|
|
end
|
2017-05-16 15:10:20 +02:00
|
|
|
end
|
|
|
|
|
2016-10-31 17:59:28 +01:00
|
|
|
[alphahat,epsilonhat,etahat,ahat,P,aK,PK,decomp,state_uncertainty] = missing_DiffuseKalmanSmootherH3_Z(ST, ...
|
2011-03-25 21:32:45 +01:00
|
|
|
Z,R1,Q,diag(H), ...
|
2014-07-15 11:13:41 +02:00
|
|
|
Pinf,Pstar,data1,vobs,np,smpl,data_index, ...
|
2015-04-03 17:48:25 +02:00
|
|
|
options_.nk,kalman_tol,diffuse_kalman_tol, ...
|
2016-10-31 17:59:28 +01:00
|
|
|
options_.filter_decomposition,options_.smoothed_state_uncertainty);
|
2011-01-13 21:50:26 +01:00
|
|
|
end
|
|
|
|
|
|
|
|
|
2016-06-16 16:25:11 +02:00
|
|
|
if expanded_state_vector_for_univariate_filter && (kalman_algo == 2 || kalman_algo == 4)
|
2011-01-13 21:50:26 +01:00
|
|
|
% extracting measurement errors
|
|
|
|
% removing observed variables from the state vector
|
2016-06-16 16:25:11 +02:00
|
|
|
k = (1:np-vobs);
|
2011-01-13 21:50:26 +01:00
|
|
|
alphahat = alphahat(k,:);
|
|
|
|
ahat = ahat(k,:);
|
|
|
|
aK = aK(:,k,:,:);
|
2016-06-16 16:25:11 +02:00
|
|
|
epsilonhat=etahat(end-vobs+1:end,:);
|
2017-05-16 15:10:20 +02:00
|
|
|
etahat=etahat(1:end-vobs,:);
|
2011-01-13 21:50:26 +01:00
|
|
|
if ~isempty(PK)
|
|
|
|
PK = PK(:,k,k,:);
|
2009-12-16 18:17:34 +01:00
|
|
|
end
|
2011-01-13 21:50:26 +01:00
|
|
|
if ~isempty(decomp)
|
|
|
|
decomp = decomp(:,k,:,:);
|
2009-12-16 18:17:34 +01:00
|
|
|
end
|
2011-01-28 14:01:57 +01:00
|
|
|
if ~isempty(P)
|
2011-01-13 21:50:26 +01:00
|
|
|
P = P(k,k,:);
|
2009-12-16 18:17:34 +01:00
|
|
|
end
|
2016-10-31 17:59:28 +01:00
|
|
|
if ~isempty(state_uncertainty)
|
|
|
|
state_uncertainty = state_uncertainty(k,k,:);
|
|
|
|
end
|
2009-12-16 18:17:34 +01:00
|
|
|
end
|