function [y_,yf_CI,yf_CI_ME]=simult_varexo_det(y0,ex,ex_det,iorder,var_list,M_,dr,exo_det_steady_state,options_) %function [y_,yf_CI,yf_CI_ME]=simult_varexo_det(y0,ex,ex_det, iorder,var_list,M_,dr,exo_det_steady_state,options_) % % Simulates a stochastic model in the presence of deterministic exogenous % shocks; allows bootstrapping standard errors for forecasting % % INPUTS: % y0: initial values, of length M_.maximum_lag % ex: matrix of stochastic exogenous shocks, starting at period 1 % ex_det: matrix of deterministic exogenous shocks, starting at period 1-M_.maximum_lag % iorder: order of approximation % var_list: list of endogenous variables to simulate % M_: Dynare model structure % dr: decision rules % exo_det_steady_state_ steady state of varexo_det % options_: Dynare options structure % % OUTPUTS: % yf: mean forecast % int_width: distance between upper bound and % mean forecast % yf_CI: upper and low bound for forecast % yf_CI_ME: upper and low bound for forecast when % considering measurement error % % % The forecast horizon is equal to size(ex, 1). % The condition size(ex,1)+M_.maximum_lag=size(ex_det,1) must be verified % for consistency. % Copyright © 2008-2025 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 . if options_.order>2 error('simult_varexo_det.m: Forecasting with varexo_det does not support order>2.') elseif options_.order==2 && options_.pruning error('simult_varexo_det.m: Forecasting with varexo_det does not support pruning.') end horizon = size(ex,1); if size(ex_det, 1) ~= horizon+M_.maximum_lag error('simult_varexo_det: Size mismatch: number of forecasting periods for stochastic exogenous and deterministic exogenous don''t match') end y_ = zeros(size(y0,1),horizon+M_.maximum_lag); y_(:,1:M_.maximum_lag) = y0; k1 = M_.maximum_lag:-1:1; %% get indices of requested variables nvar = length(var_list); if nvar == 0 nvar = M_.endo_nbr; ivar = 1:nvar; else ivar=zeros(nvar,1); for i=1:nvar i_tmp = strmatch(var_list{i}, M_.endo_names, 'exact'); if isempty(i_tmp) error ('simult_varexo_det: the requested variable %s does not exist',var_list{i}) ; else ivar(i) = i_tmp; end end end if nargout==3 % add measurement error variance uncertainty for observables [loc_H, loc_varlist] = ismember(options_.varobs, var_list); loc_varlist(loc_varlist==0) = []; %location of observables in variable list end fact = norminv((1-options_.forecasts.conf_sig)/2,0,1); if iorder == 1 for i = M_.maximum_lag+1: horizon+M_.maximum_lag tempx1 = y_(dr.order_var,k1); tempx2 = tempx1-repmat(dr.ys(dr.order_var),1,M_.maximum_lag); tempx = tempx2(M_.nstatic+(1:M_.nspred)); y_(dr.order_var,i) = dr.ys(dr.order_var)+dr.ghx*tempx+dr.ghu*ex(i-M_.maximum_lag,:)'; for j=1:min(M_.maximum_lag+dr.exo_det_length+1-i,dr.exo_det_length) y_(dr.order_var,i) = y_(dr.order_var,i) + dr.ghud{j}*(ex_det(i+j-1,:)'-exo_det_steady_state); end k1 = k1+1; end y_=y_(ivar,:); %sort according to requested var_list_ if nargout>1 % Define union of requested and state variables [var_pos]=union(dr.inv_order_var(dr.state_var),dr.inv_order_var(ivar),'stable'); % set state_pos to positions of state variables for inner part of % variance loop [~,state_pos] = ismember(dr.inv_order_var(dr.state_var),var_pos); % set var_list_pos to positions of var_list_ for reading out requested % variances [~,var_list_pos] = ismember(dr.inv_order_var(ivar),var_pos); ghx1 = dr.ghx(var_pos,:); ghu1 = dr.ghu(var_pos,:); sigma_y1 = zeros(size(ghx1,1),size(ghx1,1)); %no initial uncertainty about the states var_yf = zeros(nvar,1+horizon); %initialize for horizon_iter=1:horizon % variance recursion sigma_y1 = ghx1*sigma_y1(state_pos,state_pos)*ghx1'+ghu1*M_.Sigma_e*ghu1'; var_yf(:,1+horizon_iter) = diag(sigma_y1(var_list_pos,var_list_pos))'; % first period is initial condition without uncertainty end int_width = -fact*sqrt(var_yf); yf_CI=cat(3,y_-int_width,y_+int_width); if nargout==3 % add measurement error variance uncertainty for observables var_yf_ME=var_yf; if ~isempty(loc_varlist) var_yf_ME(loc_varlist,:) = var_yf(loc_varlist,:)+repmat(diag(M_.H(loc_H,loc_H)), 1,horizon+1); end int_width_ME = -fact*sqrt(var_yf_ME); yf_CI_ME=cat(3,y_-int_width_ME,y_+int_width_ME); %yf is already in correct order end end elseif iorder == 2 y_=simulate_varexo_det_order_2(M_,dr,y_,k1,ex,ex_det,exo_det_steady_state,horizon); if nargout>1 %% generate confidence bands via simulation % eliminate shocks with 0 variance i_exo_var = setdiff(1:M_.exo_nbr,find(diag(M_.Sigma_e) == 0)); nxs = length(i_exo_var); exo_simul = zeros(horizon,M_.exo_nbr); chol_S = chol(M_.Sigma_e(i_exo_var,i_exo_var)); y_simul=NaN(M_.endo_nbr,M_.maximum_lag+horizon,options_.forecast_replic); for forecast_iter=1:options_.forecast_replic if ~isempty(M_.Sigma_e) % we fill the shocks row wise to have the same values % independently of the length of the simulation exo_simul(:,i_exo_var) = randn(nxs,horizon)'*chol_S; end y_simul(:,:,forecast_iter) = simulate_varexo_det_order_2(M_,dr,y_(:,1:M_.maximum_lag),k1,exo_simul,ex_det,exo_det_steady_state,horizon); end yf_CI=quantile(y_simul(ivar,:,:),[(1-options_.forecasts.conf_sig)/2 1-(1-options_.forecasts.conf_sig)/2],3); end if nargout==3 % add measurement error variance uncertainty for observables if ~isempty(loc_varlist) % eliminate shocks with 0 variance i_endo_var = setdiff(1:length(options_.varobs),find(diag(M_.H) == 0)); endo_simul = zeros(horizon+1,length(options_.varobs)); chol_S = chol(M_.H(i_endo_var,i_endo_var)); for forecast_iter=1:options_.forecast_replic if ~isempty(M_.H) % we fill the shocks row wise to have the same values % independently of the length of the simulation endo_simul(:,i_endo_var,forecast_iter) = randn(length(i_endo_var),horizon+1)'*chol_S; end end y_simul(options_.varobs_id,:,:) = y_simul(options_.varobs_id,:,:)+permute(endo_simul,[2,1,3]);%overwrite old variable, not needed anymore else %no reason to add ME as no variable was requested end yf_CI_ME=quantile(y_simul(ivar,:,:),[(1-options_.forecasts.conf_sig)/2 1-(1-options_.forecasts.conf_sig)/2],3); end else error('simultxdet.m: order>2 not supported.') end function y_=simulate_varexo_det_order_2(M_,dr,y_,k1,exo,exo_det,exo_det_steady_state,horizon) for i = M_.maximum_lag+1:horizon+M_.maximum_lag tempx1 = y_(dr.order_var,k1); tempx2 = tempx1-repmat(dr.ys(dr.order_var),1,M_.maximum_lag); tempx = tempx2(M_.nstatic+(1:M_.nspred)); tempu = exo(i-M_.maximum_lag,:)'; tempuu = kron(tempu,tempu); tempxx = kron(tempx,tempx); tempxu = kron(tempx,tempu); y_(dr.order_var,i) = dr.ys(dr.order_var)+dr.ghs2/2+dr.ghx*tempx+ ... dr.ghu*tempu+0.5*(dr.ghxx*tempxx+dr.ghuu*tempuu)+dr.ghxu* ... tempxu; for j=1:min(M_.maximum_lag+dr.exo_det_length+1-i,dr.exo_det_length) tempud = exo_det(i+j-1,:)'-exo_det_steady_state; tempudud = kron(tempud,tempud); tempxud = kron(tempx,tempud); tempuud = kron(tempu,tempud); y_(dr.order_var,i) = y_(dr.order_var,i) + dr.ghud{j}*tempud + ... dr.ghxud{j}*tempxud + dr.ghuud{j}*tempuud + ... 0.5*dr.ghudud{j,j}*tempudud; for k=1:j-1 tempudk = exo_det(i+k-1,:)'-exo_det_steady_state; tempududk = kron(tempudk,tempud); y_(dr.order_var,i) = y_(dr.order_var,i) + ... dr.ghudud{k,j}*tempududk; end end k1 = k1+1; end