prior_posterior_statistics_core.m: consolidate both forecast functions into one inline function
parent
dc9560e610
commit
bf7ac27fd7
|
@ -1,67 +0,0 @@
|
|||
function yf=forcst2(y0,horizon,dr,n)
|
||||
% function yf=forcst2(y0,horizon,dr,n)
|
||||
%
|
||||
% computes forecasts based on first order model solution, given shocks
|
||||
% drawn from the shock distribution, but not including measurement error
|
||||
% Inputs:
|
||||
% - y0 [endo_nbr by maximum_endo_lag] matrix of starting values
|
||||
% - dr [structure] structure with Dynare decision rules
|
||||
% - e [horizon by exo_nbr] matrix with shocks
|
||||
% - n [scalar] number of repetitions
|
||||
%
|
||||
% Outputs:
|
||||
% - yf [horizon+ykmin_ by endo_nbr by n] array of forecasts
|
||||
%
|
||||
% Copyright © 2008-2017 Dynare Team
|
||||
%
|
||||
% 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 <https://www.gnu.org/licenses/>.
|
||||
|
||||
global M_ options_
|
||||
|
||||
Sigma_e_ = M_.Sigma_e;
|
||||
endo_nbr = M_.endo_nbr;
|
||||
exo_nbr = M_.exo_nbr;
|
||||
ykmin_ = M_.maximum_endo_lag;
|
||||
|
||||
k1 = [ykmin_:-1:1];
|
||||
k2 = dr.kstate(find(dr.kstate(:,2) <= ykmin_+1),[1 2]);
|
||||
k2 = k2(:,1)+(ykmin_+1-k2(:,2))*endo_nbr;
|
||||
|
||||
% eliminate shocks with 0 variance
|
||||
i_exo_var = setdiff([1:exo_nbr],find(diag(Sigma_e_) == 0));
|
||||
nxs = length(i_exo_var);
|
||||
|
||||
chol_S = chol(Sigma_e_(i_exo_var,i_exo_var));
|
||||
|
||||
if ~isempty(Sigma_e_)
|
||||
e = randn(nxs,n,horizon);
|
||||
end
|
||||
|
||||
B1 = dr.ghu(:,i_exo_var)*chol_S';
|
||||
|
||||
yf = zeros(endo_nbr,horizon+ykmin_,n);
|
||||
yf(:,1:ykmin_,:,:) = repmat(y0,[1,1,n]);
|
||||
|
||||
j = ykmin_*endo_nbr;
|
||||
for i=ykmin_+(1:horizon)
|
||||
tempx1 = reshape(yf(:,k1,:),[j,n]);
|
||||
tempx = tempx1(k2,:);
|
||||
yf(:,i,:) = dr.ghx*tempx+B1*squeeze(e(:,:,i-ykmin_));
|
||||
k1 = k1+1;
|
||||
end
|
||||
|
||||
yf(dr.order_var,:,:) = yf;
|
||||
yf=permute(yf,[2 1 3]);
|
|
@ -1,50 +0,0 @@
|
|||
function yf=forcst2a(y0,dr,e)
|
||||
% function yf=forcst2a(y0,dr,e)
|
||||
% computes forecasts based on first order model solution, assuming the absence of shocks
|
||||
% Inputs:
|
||||
% - y0 [endo_nbr by maximum_endo_lag] matrix of starting values
|
||||
% - dr [structure] structure with Dynare decision rules
|
||||
% - e [horizon by exo_nbr] matrix with shocks
|
||||
%
|
||||
% Outputs:
|
||||
% - yf [horizon+maximum_endo_lag,endo_nbr] matrix of forecasts
|
||||
%
|
||||
% Copyright © 2008-2017 Dynare Team
|
||||
%
|
||||
% 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 <https://www.gnu.org/licenses/>.
|
||||
|
||||
global M_ options_
|
||||
|
||||
endo_nbr = M_.endo_nbr;
|
||||
ykmin_ = M_.maximum_endo_lag;
|
||||
|
||||
horizon = size(e,1);
|
||||
|
||||
k1 = [ykmin_:-1:1];
|
||||
k2 = dr.kstate(find(dr.kstate(:,2) <= ykmin_+1),[1 2]);
|
||||
k2 = k2(:,1)+(ykmin_+1-k2(:,2))*endo_nbr;
|
||||
|
||||
yf = zeros(horizon+ykmin_,endo_nbr);
|
||||
yf(1:ykmin_,:) = y0';
|
||||
|
||||
j = ykmin_*endo_nbr;
|
||||
for i=ykmin_+(1:horizon)
|
||||
tempx = yf(k1,:)';
|
||||
yf(i,:) = tempx(k2)'*dr.ghx';
|
||||
k1 = k1+1;
|
||||
end
|
||||
|
||||
yf(:,dr.order_var) = yf;
|
|
@ -314,7 +314,7 @@ for b=fpar:B
|
|||
end
|
||||
if horizon
|
||||
yyyy = alphahat(iendo,i_last_obs);
|
||||
yf = forcst2a(yyyy,dr,zeros(horizon,exo_nbr));
|
||||
yf = simulate_posterior_forecasts(yyyy,dr,horizon,false,M_.Sigma_e,1);
|
||||
if options_.prefilter
|
||||
% add mean
|
||||
yf(:,IdObs) = yf(:,IdObs)+repmat(mean_varobs, ...
|
||||
|
@ -331,7 +331,7 @@ for b=fpar:B
|
|||
else
|
||||
yf = yf+repmat(SteadyState',horizon+maxlag,1);
|
||||
end
|
||||
yf1 = forcst2(yyyy,horizon,dr,1);
|
||||
yf1 = simulate_posterior_forecasts(yyyy,dr,horizon,false,M_.Sigma_e,1);
|
||||
if options_.prefilter == 1
|
||||
% add mean
|
||||
yf1(:,IdObs,:) = yf1(:,IdObs,:)+ ...
|
||||
|
@ -543,3 +543,56 @@ if RemoteFlag==1
|
|||
end
|
||||
|
||||
dyn_waitbar_close(h);
|
||||
|
||||
|
||||
function yf=simulate_posterior_forecasts(y0,dr,horizon,stochastic_indicator,Sigma_e,n)
|
||||
% function yf=forcst2(y0,horizon,dr,n)
|
||||
%
|
||||
% computes forecasts based on first order model solution, given shocks
|
||||
% drawn from the shock distribution, but not including measurement error
|
||||
% Inputs:
|
||||
% - y0 [endo_nbr by maximum_endo_lag] matrix of starting values
|
||||
% - dr [structure] structure with Dynare decision rules
|
||||
% - horizon [scalar] number of forecast periods
|
||||
% - stochastic_indicator [boolean] indicator whether to consider stochastic shocks
|
||||
% - Sigma_e [integer] covariance matrix of shocks
|
||||
% - n [scalar] number of repetitions
|
||||
%
|
||||
% Outputs:
|
||||
% - yf [horizon+ykmin_ by endo_nbr by n] array of forecasts
|
||||
|
||||
if nargin< 4
|
||||
stochastic_indicator=false;
|
||||
n=1;
|
||||
end
|
||||
%select states
|
||||
k2 = dr.inv_order_var(dr.state_var);
|
||||
|
||||
if stochastic_indicator
|
||||
% eliminate shocks with 0 variance
|
||||
i_exo_var = setdiff(1:length(Sigma_e),find(diag(Sigma_e) == 0));
|
||||
nxs = length(i_exo_var);
|
||||
|
||||
chol_S = chol(Sigma_e(i_exo_var,i_exo_var));
|
||||
|
||||
if ~isempty(Sigma_e)
|
||||
e = randn(nxs,n,horizon);
|
||||
end
|
||||
|
||||
B1 = dr.ghu(:,i_exo_var)*chol_S';
|
||||
end
|
||||
endo_nbr=length(y0);
|
||||
|
||||
yf = zeros(endo_nbr,1+horizon,n);
|
||||
yf(:,1,:,:) = repmat(y0,[1,1,n]);
|
||||
|
||||
for iter=1:horizon
|
||||
if stochastic_indicator
|
||||
yf(:,iter+1,:) = dr.ghx*squeeze(yf(k2,iter,:))+B1*squeeze(e(:,:,iter));
|
||||
else
|
||||
yf(:,iter+1,:) = dr.ghx*squeeze(yf(k2,iter,:));
|
||||
end
|
||||
end
|
||||
|
||||
yf(dr.order_var,:,:) = yf;
|
||||
yf=permute(yf,[2 1 3]);
|
Loading…
Reference in New Issue