| 1 |
verbatim; |
| 2 |
|
| 3 |
options_.noprint = true; |
| 4 |
|
| 5 |
number_of_simulations = 1; |
| 6 |
|
| 7 |
calibrated_values = M_.params; |
| 8 |
Sigma_e = M_.Sigma_e; |
| 9 |
|
| 10 |
options_.bnlms.set_dynare_seed_to_default = false; |
| 11 |
|
| 12 |
Beta = zeros(number_of_simulations, M_.param_nbr); |
| 13 |
|
| 14 |
for i=1:number_of_simulations |
| 15 |
% Set initial conditions randomly |
| 16 |
firstobs = rand(3, length(M_.endo_names)); |
| 17 |
% Set parameters to calibrated values (because after the |
| 18 |
% estimation parameters in equation 7 are updated with OLS |
| 19 |
% estimator). |
| 20 |
M_.params = calibrated_values; |
| 21 |
M_.Sigma_e = Sigma_e; |
| 22 |
% Simulate the model. |
| 23 |
simdata = simul_backward_model(dseries(firstobs, dates('1995Q1'), M_.endo_names), 1100); |
| 24 |
% Select a subsample. |
| 25 |
simdata = simdata(simdata.dates(101:1100)); |
| 26 |
% Get vector of indices to exclude exogenous variables from the dataset. |
| 27 |
names = regexp(simdata.name, 'res\w*'); |
| 28 |
idxs = find(cellfun(@isempty, names)); |
| 29 |
% Perform the estimation of equation 7. |
| 30 |
if strcmp(M_.fname, 'ols_param_names') |
| 31 |
dyn_ols(simdata{idxs}, {}, {'eq7'}, '', {{'de_g_yer_ecm_u2_stn_L1', 'de_g_yer_de_g_yer_L1'}}); |
| 32 |
elseif strcmp(M_.fname, 'ols_date_range') |
| 33 |
dyn_ols(simdata{idxs}, {}, {'eq7'}, '', {}, simdata.dates(2:end-3)); |
| 34 |
else |
| 35 |
dyn_ols(simdata{idxs}, {}, {'eq7'}); |
| 36 |
end |
| 37 |
if i ~= number_of_simulations |
| 38 |
oo_ = rmfield(oo_, 'ols'); |
| 39 |
end |
| 40 |
% Store the estimation results in Beta |
| 41 |
Beta(i, :) = M_.params'; |
| 42 |
end |
| 43 |
|
| 44 |
% Get the indices of the estimated parameters |
| 45 |
pid = oo_.ols.eq7.param_idxs; |
| 46 |
|
| 47 |
if number_of_simulations > 1 |
| 48 |
if max(abs(calibrated_values(pid)-mean(Beta(:,pid)', 2)))>1e-2 |
| 49 |
error('There is probably an error in the OLS routine.') |
| 50 |
end |
| 51 |
else |
| 52 |
% NB: estimation results depend on the sequence of generated random |
| 53 |
% numbers, hence we need to hardcode different results between MATLAB |
| 54 |
% and Octave |
| 55 |
if strcmp(M_.fname, 'ols_base') |
| 56 |
if isoctave |
| 57 |
good = [2.509401520579608e-02 |
| 58 |
-1.212468793686628e-01 |
| 59 |
2.983003417202038e-02 |
| 60 |
4.645373135672548e-01 |
| 61 |
-6.312448906643650e-02]; |
| 62 |
else |
| 63 |
good = [0.025487074270635 |
| 64 |
-0.126705792250240 |
| 65 |
0.002639663639062 |
| 66 |
0.486913131141265 |
| 67 |
-0.027702100375691]; |
| 68 |
end |
| 69 |
elseif strcmp(M_.fname, 'ols_param_names') |
| 70 |
if isoctave |
| 71 |
good = [-0.125307512725951 |
| 72 |
0.465364467369817]; |
| 73 |
else |
| 74 |
good = [-0.125265361034515 |
| 75 |
0.497003481797959]; |
| 76 |
end |
| 77 |
elseif strcmp(M_.fname, 'ols_date_range') |
| 78 |
if isoctave |
| 79 |
good = [4.775810440061960e-01 |
| 80 |
2.096533656963026e-02 |
| 81 |
3.264889547460539e-02 |
| 82 |
-1.167420901882959e-01]; |
| 83 |
else |
| 84 |
good = [0.468607715487806 |
| 85 |
-0.011224442004140 |
| 86 |
0.033681224841824 |
| 87 |
-0.126743281213504]; |
| 88 |
end |
| 89 |
elseif strcmp(M_.fname, 'ols_wc_1') |
| 90 |
if isoctave |
| 91 |
good = [4.778519485900916e-01 |
| 92 |
2.028231364527827e-02 |
| 93 |
3.186766890724485e-02 |
| 94 |
-1.173897445629599e-01]; |
| 95 |
else |
| 96 |
good = [0.466534186307255 |
| 97 |
-0.010596174928406 |
| 98 |
0.035098464382347 |
| 99 |
-0.126502172176703]; |
| 100 |
end |
| 101 |
elseif strcmp(M_.fname, 'ols_wc_2') |
| 102 |
if isoctave |
| 103 |
good = [4.679679703340907e-01 |
| 104 |
2.987618894535485e-02 |
| 105 |
2.503849196608867e-02 |
| 106 |
-1.206391201296474e-01 |
| 107 |
3.435379133407408e-02]; |
| 108 |
else |
| 109 |
good = [0.485802603276717 |
| 110 |
0.002660854661647 |
| 111 |
0.025434782170432 |
| 112 |
-0.126939890810119 |
| 113 |
0.043688440124315]; |
| 114 |
end |
| 115 |
elseif strcmp(M_.fname, 'ols_wc_3') |
| 116 |
if isoctave |
| 117 |
good = [4.766230170524117e-01 |
| 118 |
2.500479091401997e-02 |
| 119 |
2.333265887171398e-02 |
| 120 |
-9.913848386063488e-02]; |
| 121 |
else |
| 122 |
good = [0.490636551206316 |
| 123 |
0.014209413280418 |
| 124 |
0.028534044956486 |
| 125 |
-0.094387460262917]; |
| 126 |
end |
| 127 |
end |
| 128 |
if max(abs(good-Beta(:,pid)'))>1e-2 |
| 129 |
error('There is probably an error in the OLS routine.') |
| 130 |
end |
| 131 |
return |
| 132 |
end |
| 133 |
|
| 134 |
for i=1:length(pid) |
| 135 |
figure(i) |
| 136 |
hold on |
| 137 |
title(strrep(M_.param_names(pid(i),:), '_', '\_')); |
| 138 |
bandwidth = mh_optimal_bandwidth(Beta(:,pid(i)), length(Beta(:,pid(i))), -1, 'gaussian'); |
| 139 |
[abscissa, f] = kernel_density_estimate(Beta(:,pid(i)), 256, length(Beta(:,pid(i))), bandwidth, 'gaussian'); |
| 140 |
plot(abscissa, f, '-k', 'linewidth', 2); |
| 141 |
line([calibrated_values(pid(i)) calibrated_values(pid(i))], [0 max(f)*1.05], 'LineWidth', 2, 'Color', 'r'); |
| 142 |
line([mean(Beta(:,pid(i))) mean(Beta(:,pid(i)))], [0 max(f)*1.05], 'LineWidth', 2, 'Color', 'g'); |
| 143 |
axis tight |
| 144 |
box on |
| 145 |
hold off |
| 146 |
end |
| 147 |
|
| 148 |
end; |
| 149 |
|