Add unit tests for HP filter, bandpass filter, and spectral density
parent
40de494568
commit
2c63ca8843
|
@ -7,7 +7,10 @@ MODFILES = \
|
|||
estimation/MH_recover/fs2000_recover.mod \
|
||||
estimation/t_proposal/fs2000_student.mod \
|
||||
estimation/TaRB/fs2000_tarb.mod \
|
||||
moments/example1_var_decomp \
|
||||
moments/example1_var_decomp.mod \
|
||||
moments/example1_hp_test.mod \
|
||||
moments/example1_bp_test.mod \
|
||||
moments/test_AR1_spectral_density.mod \
|
||||
gsa/ls2003.mod \
|
||||
gsa/ls2003a.mod \
|
||||
gsa/cod_ML_morris/cod_ML_morris.mod \
|
||||
|
|
|
@ -0,0 +1,116 @@
|
|||
/*
|
||||
* Example 1 from F. Collard (2001): "Stochastic simulations with DYNARE:
|
||||
* A practical guide" (see "guide.pdf" in the documentation directory).
|
||||
*/
|
||||
|
||||
/*
|
||||
* Copyright (C) 2001-2010 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 <http://www.gnu.org/licenses/>.
|
||||
*/
|
||||
|
||||
|
||||
var y, c, k, a, h, b;
|
||||
varexo e, u;
|
||||
|
||||
parameters beta, rho, alpha, delta, theta, psi, tau;
|
||||
|
||||
alpha = 0.36;
|
||||
rho = 0.95;
|
||||
tau = 0.025;
|
||||
beta = 0.99;
|
||||
delta = 0.025;
|
||||
psi = 0;
|
||||
theta = 2.95;
|
||||
|
||||
phi = 0.1;
|
||||
|
||||
model;
|
||||
c*theta*h^(1+psi)=(1-alpha)*y;
|
||||
k = beta*(((exp(b)*c)/(exp(b(+1))*c(+1)))
|
||||
*(exp(b(+1))*alpha*y(+1)+(1-delta)*k));
|
||||
y = exp(a)*(k(-1)^alpha)*(h^(1-alpha));
|
||||
k = exp(b)*(y-c)+(1-delta)*k(-1);
|
||||
a = rho*a(-1)+tau*b(-1) + e;
|
||||
b = tau*a(-1)+rho*b(-1) + u;
|
||||
end;
|
||||
|
||||
initval;
|
||||
y = 1.08068253095672;
|
||||
c = 0.80359242014163;
|
||||
h = 0.29175631001732;
|
||||
k = 11.08360443260358;
|
||||
a = 0;
|
||||
b = 0;
|
||||
e = 0;
|
||||
u = 0;
|
||||
end;
|
||||
|
||||
shocks;
|
||||
var e; stderr 0.009;
|
||||
var u; stderr 0.009;
|
||||
var e, u = phi*0.009*0.009;
|
||||
end;
|
||||
|
||||
steady(solve_algo=4,maxit=1000);
|
||||
|
||||
stoch_simul(order=1,nofunctions,irf=0,bandpass_filter=[6 32],hp_ngrid=8192);
|
||||
oo_filtered_all_shocks_theoretical=oo_;
|
||||
|
||||
stoch_simul(order=1,nofunctions,periods=1000000);
|
||||
oo_filtered_all_shocks_simulated=oo_;
|
||||
|
||||
|
||||
shocks;
|
||||
var e; stderr 0;
|
||||
var u; stderr 0.009;
|
||||
var e, u = phi*0.009*0;
|
||||
end;
|
||||
|
||||
stoch_simul(order=1,nofunctions,irf=0,periods=0);
|
||||
|
||||
oo_filtered_one_shock_theoretical=oo_;
|
||||
|
||||
stoch_simul(order=1,nofunctions,periods=5000000);
|
||||
oo_filtered_one_shock_simulated=oo_;
|
||||
|
||||
|
||||
if max(abs((1-diag(oo_filtered_one_shock_simulated.var)./(diag(oo_filtered_all_shocks_simulated.var)))*100-oo_filtered_all_shocks_theoretical.variance_decomposition(:,1)))>2
|
||||
error('Variance Decomposition wrong')
|
||||
end
|
||||
|
||||
if max(max(abs(oo_filtered_all_shocks_simulated.var-oo_filtered_all_shocks_theoretical.var)))>2e-4;
|
||||
error('Covariance wrong')
|
||||
end
|
||||
|
||||
if max(max(abs(oo_filtered_one_shock_simulated.var-oo_filtered_one_shock_theoretical.var)))>1e-4;
|
||||
error('Covariance wrong')
|
||||
end
|
||||
|
||||
for ii=1:options_.ar
|
||||
autocorr_model_all_shocks_simulated(:,ii)=diag(oo_filtered_all_shocks_simulated.autocorr{ii});
|
||||
autocorr_model_all_shocks_theoretical(:,ii)=diag(oo_filtered_all_shocks_theoretical.autocorr{ii});
|
||||
autocorr_model_one_shock_simulated(:,ii)=diag(oo_filtered_one_shock_simulated.autocorr{ii});
|
||||
autocorr_model_one_shock_theoretical(:,ii)=diag(oo_filtered_one_shock_theoretical.autocorr{ii});
|
||||
end
|
||||
|
||||
if max(max(abs(autocorr_model_all_shocks_simulated-autocorr_model_all_shocks_theoretical)))>2e-2;
|
||||
error('Correlation wrong')
|
||||
end
|
||||
|
||||
if max(max(abs(autocorr_model_one_shock_simulated-autocorr_model_one_shock_theoretical)))>2e-2;
|
||||
error('Correlation wrong')
|
||||
end
|
|
@ -0,0 +1,157 @@
|
|||
/*
|
||||
* Example 1 from F. Collard (2001): "Stochastic simulations with DYNARE:
|
||||
* A practical guide" (see "guide.pdf" in the documentation directory).
|
||||
*/
|
||||
|
||||
/*
|
||||
* Copyright (C) 2001-2010 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 <http://www.gnu.org/licenses/>.
|
||||
*/
|
||||
|
||||
|
||||
var y, c, k, a, h, b;
|
||||
varexo e, u;
|
||||
|
||||
parameters beta, rho, alpha, delta, theta, psi, tau;
|
||||
|
||||
alpha = 0.36;
|
||||
rho = 0.95;
|
||||
tau = 0.025;
|
||||
beta = 0.99;
|
||||
delta = 0.025;
|
||||
psi = 0;
|
||||
theta = 2.95;
|
||||
|
||||
phi = 0.1;
|
||||
|
||||
model;
|
||||
c*theta*h^(1+psi)=(1-alpha)*y;
|
||||
k = beta*(((exp(b)*c)/(exp(b(+1))*c(+1)))
|
||||
*(exp(b(+1))*alpha*y(+1)+(1-delta)*k));
|
||||
y = exp(a)*(k(-1)^alpha)*(h^(1-alpha));
|
||||
k = exp(b)*(y-c)+(1-delta)*k(-1);
|
||||
a = rho*a(-1)+tau*b(-1) + e;
|
||||
b = tau*a(-1)+rho*b(-1) + u;
|
||||
end;
|
||||
|
||||
initval;
|
||||
y = 1.08068253095672;
|
||||
c = 0.80359242014163;
|
||||
h = 0.29175631001732;
|
||||
k = 11.08360443260358;
|
||||
a = 0;
|
||||
b = 0;
|
||||
e = 0;
|
||||
u = 0;
|
||||
end;
|
||||
|
||||
shocks;
|
||||
var e; stderr 0.009;
|
||||
var u; stderr 0.009;
|
||||
var e, u = phi*0.009*0.009;
|
||||
end;
|
||||
|
||||
steady(solve_algo=4,maxit=1000);
|
||||
options_.hp_ngrid=2048*4;
|
||||
options_.bandpass.indicator=0;
|
||||
options_.bandpass.passband=[6 32];
|
||||
|
||||
stoch_simul(order=1,nofunctions,hp_filter=1600,irf=0);
|
||||
|
||||
total_var_filtered=diag(oo_.var);
|
||||
oo_filtered_all_shocks=oo_;
|
||||
|
||||
stoch_simul(order=1,nofunctions,hp_filter=0,periods=5000000,nomoments);
|
||||
options_.nomoments=0;
|
||||
oo_unfiltered_all_shocks=oo_;
|
||||
|
||||
[junk, y_filtered]=sample_hp_filter(y,1600);
|
||||
[junk, c_filtered]=sample_hp_filter(c,1600);
|
||||
[junk, k_filtered]=sample_hp_filter(k,1600);
|
||||
[junk, a_filtered]=sample_hp_filter(a,1600);
|
||||
[junk, h_filtered]=sample_hp_filter(h,1600);
|
||||
[junk, b_filtered]=sample_hp_filter(b,1600);
|
||||
|
||||
verbatim;
|
||||
total_std_all_shocks_filtered_sim=std([y_filtered c_filtered k_filtered a_filtered h_filtered b_filtered])
|
||||
cov_filtered_all_shocks=cov([y_filtered c_filtered k_filtered a_filtered h_filtered b_filtered])
|
||||
acf(1,:)=autocorr([y_filtered ],5)';
|
||||
acf(2,:)=autocorr([c_filtered ],5)';
|
||||
acf(3,:)=autocorr([k_filtered ],5)';
|
||||
acf(4,:)=autocorr([a_filtered ],5)';
|
||||
acf(5,:)=autocorr([h_filtered ],5)';
|
||||
acf(6,:)=autocorr([b_filtered ],5)';
|
||||
autocorr_filtered_all_shocks=acf(:,2:end);
|
||||
end;
|
||||
|
||||
shocks;
|
||||
var e; stderr 0;
|
||||
var u; stderr 0.009;
|
||||
var e, u = phi*0.009*0;
|
||||
end;
|
||||
|
||||
stoch_simul(order=1,nofunctions,hp_filter=1600,irf=0,periods=0);
|
||||
|
||||
total_var_filtered_one_shock=diag(oo_.var);
|
||||
oo_filtered_one_shock=oo_;
|
||||
|
||||
stoch_simul(order=1,nofunctions,hp_filter=0,periods=5000000,nomoments);
|
||||
oo_unfiltered_one_shock=oo_;
|
||||
|
||||
[junk, y_filtered]=sample_hp_filter(y,1600);
|
||||
[junk, c_filtered]=sample_hp_filter(c,1600);
|
||||
[junk, k_filtered]=sample_hp_filter(k,1600);
|
||||
[junk, a_filtered]=sample_hp_filter(a,1600);
|
||||
[junk, h_filtered]=sample_hp_filter(h,1600);
|
||||
[junk, b_filtered]=sample_hp_filter(b,1600);
|
||||
|
||||
verbatim;
|
||||
total_std_one_shock_filtered_sim=std([y_filtered c_filtered k_filtered a_filtered h_filtered b_filtered])
|
||||
cov_filtered_one_shock=cov([y_filtered c_filtered k_filtered a_filtered h_filtered b_filtered])
|
||||
acf(1,:)=autocorr([y_filtered ],5)';
|
||||
acf(2,:)=autocorr([c_filtered ],5)';
|
||||
acf(3,:)=autocorr([k_filtered ],5)';
|
||||
acf(4,:)=autocorr([a_filtered ],5)';
|
||||
acf(5,:)=autocorr([h_filtered ],5)';
|
||||
acf(6,:)=autocorr([b_filtered ],5)';
|
||||
autocorr_filtered_one_shock=acf(:,2:end);
|
||||
end;
|
||||
|
||||
if max(abs((1-(total_std_one_shock_filtered_sim.^2)./(total_std_all_shocks_filtered_sim.^2))*100-oo_filtered_all_shocks.variance_decomposition(:,1)'))>2
|
||||
error('Variance Decomposition wrong')
|
||||
end
|
||||
|
||||
if max(max(abs(oo_filtered_all_shocks.var-cov_filtered_all_shocks)))>1e-4;
|
||||
error('Covariance wrong')
|
||||
end
|
||||
|
||||
if max(max(abs(oo_filtered_one_shock.var-cov_filtered_one_shock)))>5e-5;
|
||||
error('Covariance wrong')
|
||||
end
|
||||
|
||||
for ii=1:options_.ar
|
||||
autocorr_model_all_shocks(:,ii)=diag(oo_filtered_all_shocks.autocorr{ii});
|
||||
autocorr_model_one_shock(:,ii)=diag(oo_filtered_one_shock.autocorr{ii});
|
||||
end
|
||||
|
||||
if max(max(abs(autocorr_model_all_shocks-autocorr_filtered_all_shocks)))>1e-2;
|
||||
error('Covariance wrong')
|
||||
end
|
||||
|
||||
if max(max(abs(autocorr_model_one_shock-autocorr_filtered_one_shock)))>1e-2;
|
||||
error('Covariance wrong')
|
||||
end
|
|
@ -0,0 +1,76 @@
|
|||
var white_noise ar1;
|
||||
varexo e;
|
||||
|
||||
parameters phi;
|
||||
|
||||
phi=0.9;
|
||||
|
||||
model;
|
||||
white_noise=e;
|
||||
ar1=phi*ar1(-1)+e;
|
||||
|
||||
end;
|
||||
|
||||
shocks;
|
||||
var e = 1;
|
||||
end;
|
||||
|
||||
options_.SpectralDensity.trigger=1;
|
||||
|
||||
options_.bandpass.indicator=0;
|
||||
options_.hp_ngrid=2048;
|
||||
|
||||
stoch_simul(order=1,nofunctions,hp_filter=0,irf=0,periods=1000000);
|
||||
|
||||
white_noise_sample=white_noise;
|
||||
|
||||
theoretical_spectrum_white_noise=1^2/(2*pi); %Hamilton (1994), 6.1.9
|
||||
if max(abs(oo_.SpectralDensity.density(strmatch('white_noise',M_.endo_names,'exact'),:)-theoretical_spectrum_white_noise))>1e-10
|
||||
error('Spectral Density is wrong')
|
||||
end
|
||||
|
||||
theoretical_spectrum_AR1=1/(2*pi)*(1^2./(1+phi^2-2*phi*cos(oo_.SpectralDensity.freqs))); %Hamilton (1994), 6.1.13
|
||||
if max(abs(oo_.SpectralDensity.density(strmatch('ar1',M_.endo_names,'exact'),:)-theoretical_spectrum_AR1'))>1e-10
|
||||
error('Spectral Density is wrong')
|
||||
end
|
||||
|
||||
stoch_simul(order=1,nofunctions,hp_filter=1600,irf=0,periods=0);
|
||||
lambda=options_.hp_filter;
|
||||
Kalman_gain=(4*lambda*(1 - cos(oo_.SpectralDensity.freqs)).^2 ./ (1 + 4*lambda*(1 - cos(oo_.SpectralDensity.freqs)).^2));
|
||||
theoretical_spectrum_white_noise_hp_filtered=1^2/(2*pi)*Kalman_gain.^2; %Hamilton (1994), 6.1.9
|
||||
if max(abs(oo_.SpectralDensity.density(strmatch('white_noise',M_.endo_names,'exact'),:)-theoretical_spectrum_white_noise_hp_filtered'))>1e-10
|
||||
error('Spectral Density is wrong')
|
||||
end
|
||||
|
||||
theoretical_spectrum_AR1_hp_filtered=1/(2*pi)*(1^2./(1+phi^2-2*phi*cos(oo_.SpectralDensity.freqs))).*Kalman_gain.^2; %Hamilton (1994), 6.1.13
|
||||
if max(abs(oo_.SpectralDensity.density(strmatch('ar1',M_.endo_names,'exact'),:)-theoretical_spectrum_AR1_hp_filtered'))>1e-10
|
||||
error('Spectral Density is wrong')
|
||||
end
|
||||
|
||||
options_.hp_filter=0;
|
||||
stoch_simul(order=1,nofunctions,bandpass_filter=[6 32],irf=0);
|
||||
|
||||
theoretical_spectrum_white_noise=repmat(theoretical_spectrum_white_noise,1,options_.hp_ngrid);
|
||||
passband=oo_.SpectralDensity.freqs>=2*pi/options_.bandpass.passband(2) & oo_.SpectralDensity.freqs<=2*pi/options_.bandpass.passband(1);
|
||||
if max(abs(oo_.SpectralDensity.density(strmatch('white_noise',M_.endo_names,'exact'),passband)-theoretical_spectrum_white_noise(passband)))>1e-10
|
||||
error('Spectral Density is wrong')
|
||||
end
|
||||
if max(abs(oo_.SpectralDensity.density(strmatch('white_noise',M_.endo_names,'exact'),~passband)-0))>1e-10
|
||||
error('Spectral Density is wrong')
|
||||
end
|
||||
|
||||
if max(abs(oo_.SpectralDensity.density(strmatch('ar1',M_.endo_names,'exact'),passband)-theoretical_spectrum_AR1(passband)'))>1e-10
|
||||
error('Spectral Density is wrong')
|
||||
end
|
||||
if max(abs(oo_.SpectralDensity.density(strmatch('ar1',M_.endo_names,'exact'),~passband)-0))>1e-10
|
||||
error('Spectral Density is wrong')
|
||||
end
|
||||
|
||||
|
||||
% [pow,f]=psd(a_sample,1024,1,[],512);
|
||||
% figure
|
||||
% plot(f,pow/(2*pi))
|
||||
%
|
||||
% % figure
|
||||
% % [pow,f]=psd(a_sample,1000,1,[],500);
|
||||
% % plot(f(3:end)*2*pi,pow(3:end)/(2*pi));
|
Loading…
Reference in New Issue