| 1 |
% DSGE model used in replication files of |
| 2 |
% An, Sungbae and Schorfheide, Frank, (2007), Bayesian Analysis of DSGE Models, Econometric Reviews, 26, issue 2-4, p. 113-172. |
| 3 |
% Adapted by Willi Mutschler (@wmutschl, [email protected]) |
| 4 |
% ========================================================================= |
| 5 |
% Copyright © 2020-2021 Dynare Team |
| 6 |
% |
| 7 |
% This file is part of Dynare. |
| 8 |
% |
| 9 |
% Dynare is free software: you can redistribute it and/or modify |
| 10 |
% it under the terms of the GNU General Public License as published by |
| 11 |
% the Free Software Foundation, either version 3 of the License, or |
| 12 |
% (at your option) any later version. |
| 13 |
% |
| 14 |
% Dynare is distributed in the hope that it will be useful, |
| 15 |
% but WITHOUT ANY WARRANTY; without even the implied warranty of |
| 16 |
% MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
| 17 |
% GNU General Public License for more details. |
| 18 |
% |
| 19 |
% You should have received a copy of the GNU General Public License |
| 20 |
% along with Dynare. If not, see <https://www.gnu.org/licenses/>. |
| 21 |
% ========================================================================= |
| 22 |
|
| 23 |
var c p R g y z INFL INT YGR; |
| 24 |
varexo e_r e_g e_z; |
| 25 |
parameters tau nu kap cyst psi1 psi2 rhor rhog rhoz rrst pist gamst; |
| 26 |
|
| 27 |
varobs INT YGR INFL; |
| 28 |
|
| 29 |
tau = 2; |
| 30 |
nu = 0.1; |
| 31 |
kap = 0.33; |
| 32 |
cyst = 0.85; |
| 33 |
psi1 = 1.5; |
| 34 |
psi2 = 0.125; |
| 35 |
rhor = 0.75; |
| 36 |
rhog = 0.95; |
| 37 |
rhoz = 0.9; |
| 38 |
rrst = 1; |
| 39 |
pist = 3.2; |
| 40 |
gamst = 0.55; |
| 41 |
|
| 42 |
model; |
| 43 |
#pist2 = exp(pist/400); |
| 44 |
#rrst2 = exp(rrst/400); |
| 45 |
#bet = 1/rrst2; |
| 46 |
#phi = tau*(1-nu)/nu/kap/pist2^2; |
| 47 |
#gst = 1/cyst; |
| 48 |
#cst = (1-nu)^(1/tau); |
| 49 |
#yst = cst*gst; |
| 50 |
#dy = y-y(-1); |
| 51 |
1 = exp(-tau*c(+1)+tau*c+R-z(+1)-p(+1)); |
| 52 |
(1-nu)/nu/phi/(pist2^2)*(exp(tau*c)-1) = (exp(p)-1)*((1-1/2/nu)*exp(p)+1/2/nu) - bet*(exp(p(+1))-1)*exp(-tau*c(+1)+tau*c+y(+1)-y+p(+1)); |
| 53 |
exp(c-y) = exp(-g) - phi*pist2^2*gst/2*(exp(p)-1)^2; |
| 54 |
R = rhor*R(-1) + (1-rhor)*psi1*p + (1-rhor)*psi2*(dy+z) + e_r/100; |
| 55 |
g = rhog*g(-1) + e_g/100; |
| 56 |
z = rhoz*z(-1) + e_z/100; |
| 57 |
YGR = gamst+100*(dy+z); |
| 58 |
INFL = pist+400*p; |
| 59 |
INT = pist+rrst+4*gamst+400*R; |
| 60 |
end; |
| 61 |
|
| 62 |
steady_state_model; |
| 63 |
z = 0; p = 0; g = 0; R = 0; c = 0; y = 0; |
| 64 |
YGR = gamst; INFL = pist; INT = pist + rrst + 4*gamst; |
| 65 |
end; |
| 66 |
|
| 67 |
shocks; |
| 68 |
var e_r = 0.20^2; |
| 69 |
var e_g = 0.80^2; |
| 70 |
var e_z = 0.45^2; |
| 71 |
corr e_r,e_g = 0.2; |
| 72 |
end; |
| 73 |
|
| 74 |
@#if estimParams == 0 |
| 75 |
% Define only initial values without bounds |
| 76 |
estimated_params; |
| 77 |
%tau, 1.50; |
| 78 |
%kap, 0.15; |
| 79 |
psi1, 1.20; |
| 80 |
psi2, 0.50; |
| 81 |
rhor, 0.50; |
| 82 |
%rhog, 0.50; |
| 83 |
%rhoz, 0.50; |
| 84 |
%rrst, 1.20; |
| 85 |
%pist, 3.00; |
| 86 |
gamst, 0.75; |
| 87 |
stderr e_r, 0.30; |
| 88 |
stderr e_g, 0.30; |
| 89 |
stderr e_z, 0.30; |
| 90 |
corr e_r,e_g, 0.10; |
| 91 |
end; |
| 92 |
@#endif |
| 93 |
|
| 94 |
@#if estimParams == 1 |
| 95 |
% Define initial values and bounds |
| 96 |
estimated_params; |
| 97 |
%tau, 1.50, 1e-5, 10; |
| 98 |
%kap, 0.15, 1e-5, 10; |
| 99 |
psi1, 1.20, 1e-5, 10; |
| 100 |
psi2, 0.50, 1e-5, 10; |
| 101 |
rhor, 0.50, 1e-5, 0.99999; |
| 102 |
%rhog, 0.50, 1e-5, 0.99999; |
| 103 |
%rhoz, 0.50, 1e-5, 0.99999; |
| 104 |
%rrst, 1.20, 1e-5, 10; |
| 105 |
%pist, 3.00, 1e-5, 20; |
| 106 |
gamst, 0.75, -5, 5; |
| 107 |
stderr e_r, 0.30, 1e-8, 5; |
| 108 |
stderr e_g, 0.30, 1e-8, 5; |
| 109 |
stderr e_z, 0.30, 1e-8, 5; |
| 110 |
corr e_r,e_g, 0.10, -1, 1; |
| 111 |
end; |
| 112 |
@#endif |
| 113 |
|
| 114 |
@#if estimParams == 2 |
| 115 |
% Define prior distribution |
| 116 |
estimated_params; |
| 117 |
%tau, 1.50, 1e-5, 10, gamma_pdf, 2.00, 0.50; |
| 118 |
%kap, 0.15, 1e-5, 10, gamma_pdf, 0.33, 0.10; |
| 119 |
psi1, 1.20, 1e-5, 10, gamma_pdf, 1.50, 0.25; |
| 120 |
psi2, 0.50, 1e-5, 10, gamma_pdf, 0.125, 0.25; |
| 121 |
rhor, 0.50, 1e-5, 0.99999, beta_pdf, 0.50, 0.20; |
| 122 |
%rhog, 0.50, 1e-5, 0.99999, beta_pdf, 0.80, 0.10; |
| 123 |
%rhoz, 0.50, 1e-5, 0.99999, beta_pdf, 0.66, 0.15; |
| 124 |
%rrst, 1.20, 1e-5, 10, gamma_pdf, 0.50, 0.50; |
| 125 |
%pist, 3.00, 1e-5, 20, gamma_pdf, 7.00, 2.00; |
| 126 |
gamst, 0.75, -5, 5, normal_pdf, 0.40, 0.20; |
| 127 |
stderr e_r, 0.30, 1e-8, 5, inv_gamma_pdf, 0.50, 0.26; |
| 128 |
stderr e_g, 0.30, 1e-8, 5, inv_gamma_pdf, 1.25, 0.65; |
| 129 |
stderr e_z, 0.30, 1e-8, 5, inv_gamma_pdf, 0.63, 0.33; |
| 130 |
corr e_r,e_g, 0.10, -1, 1, uniform_pdf, , , -1, 1; |
| 131 |
end; |
| 132 |
@#endif |
| 133 |
|
| 134 |
|
| 135 |
% % Simulate data |
| 136 |
% stoch_simul(order=2,pruning,nodisplay,nomoments,periods=750,drop=500); |
| 137 |
% save('AnScho_MoM_data_2.mat', options_.varobs{:} ); |
| 138 |
% pause(1); |
| 139 |
|
| 140 |
%-------------------------------------------------------------------------- |
| 141 |
% Method of Moments Estimation |
| 142 |
%-------------------------------------------------------------------------- |
| 143 |
matched_moments; |
| 144 |
|
| 145 |
%first-order product moments |
| 146 |
YGR; |
| 147 |
YGR(-1); %redundant |
| 148 |
YGR(0)^1; %redundant |
| 149 |
YGR^1; %redundant |
| 150 |
YGR^1*INFL^0*INT^0; %redundant |
| 151 |
INFL(+2)^1; |
| 152 |
INT(-2); |
| 153 |
INT(2); %redundant |
| 154 |
|
| 155 |
%second-order contemporenous product moments |
| 156 |
YGR^2; |
| 157 |
YGR(-1)^2; %redundant |
| 158 |
YGR*YGR; %redundant |
| 159 |
YGR(1)*YGR(1); %redundant |
| 160 |
INFL*YGR; |
| 161 |
YGR*INFL; %redundant |
| 162 |
YGR(2)*INT(2); |
| 163 |
INT(2)*YGR(2); %redundant |
| 164 |
INT(2)^1*YGR(2)^1; %redundant |
| 165 |
YGR(2)^1*INT(2)^1; %redundant |
| 166 |
INFL(-2)^1*INFL(-2)^1; |
| 167 |
INFL(-1)*INT(-1); |
| 168 |
INT(0)^1*INT^1; |
| 169 |
INT(0)^1*INT^1*YGR(2)^0*INT(-2)^0*INFL^0; %redundant |
| 170 |
|
| 171 |
%second-order temporal product moments |
| 172 |
YGR*YGR(-1); |
| 173 |
YGR(3)*YGR(2); %redundant |
| 174 |
YGR(2)*YGR(3); %redundant |
| 175 |
YGR(-2)*YGR(-1); %redundant |
| 176 |
INT(3)^1*INT(-2)^1; |
| 177 |
YGR(5)^0*INFL*INFL(2)^1; |
| 178 |
YGR(5)^1*INFL(-3)^1; |
| 179 |
INFL(-3)^1*YGR(5)^1; %redundant |
| 180 |
INFL(3)*INT(2); |
| 181 |
|
| 182 |
@#ifdef Extended_Matched_Moments_Checks |
| 183 |
YGR^3; |
| 184 |
YGR(-3)^2*YGR(-3); %redundant |
| 185 |
YGR(-3)*YGR(-3)^2; %redundant |
| 186 |
YGR(0)^1*YGR(0)*YGR; %redundant |
| 187 |
INT(-2)^2*INFL(-1)^4; |
| 188 |
INFL(1)^4*INT(0)^2; %redundant |
| 189 |
INT(0)^2*INFL(1)^4; %redundant |
| 190 |
%YGR^2*INFL(-3)^4*INT(5)^6; |
| 191 |
%YGR^2*INT(5)^6*INFL(-3)^4;%redundant |
| 192 |
%INT(5)^6*YGR^2*INFL(-3)^4;%redundant |
| 193 |
@#endif |
| 194 |
end; |
| 195 |
|
| 196 |
M_.matched_moments_orig = M_.matched_moments; |
| 197 |
|
| 198 |
method_of_moments( |
| 199 |
% Necessary options |
| 200 |
mom_method = @{MoM_Method} % method of moments method; possible values: GMM|SMM |
| 201 |
, datafile = 'AnScho_MoM_data_2.mat' % name of filename with data |
| 202 |
|
| 203 |
% Options for both GMM and SMM |
| 204 |
% , bartlett_kernel_lag = 20 % bandwith in optimal weighting matrix |
| 205 |
, order = 2 % order of Taylor approximation in perturbation |
| 206 |
% , penalized_estimator % include deviation from prior mean as additional moment restriction and use prior precision as weight |
| 207 |
, pruning % use pruned state space system at higher-order |
| 208 |
, verbose % display and store intermediate estimation results |
| 209 |
, weighting_matrix = ['optimal'] % weighting matrix in moments distance objective function; possible values: OPTIMAL|IDENTITY_MATRIX|DIAGONAL|filename. Size of cell determines stages in iterated estimation, e.g. two state with ['DIAGONAL','OPTIMAL'] |
| 210 |
%, weighting_matrix_scaling_factor=1 % scaling of weighting matrix in objective function |
| 211 |
, se_tolx=1e-6 % step size for numerical computation of standard errors |
| 212 |
|
| 213 |
% Options for SMM |
| 214 |
, burnin=50 % number of periods dropped at beginning of simulation |
| 215 |
, bounded_shock_support % trim shocks in simulation to +- 2 stdev |
| 216 |
% , seed = 24051986 % seed used in simulations |
| 217 |
, simulation_multiple = 5 % multiple of the data length used for simulation |
| 218 |
|
| 219 |
% Options for GMM |
| 220 |
@#if MoM_Method == "GMM" |
| 221 |
, analytic_standard_errors % compute standard errors using analytical derivatives |
| 222 |
@#endif |
| 223 |
% General options |
| 224 |
% , dirname = 'MM' % directory in which to store estimation output |
| 225 |
% , graph_format = EPS % specify the file format(s) for graphs saved to disk |
| 226 |
% , nodisplay % do not display the graphs, but still save them to disk |
| 227 |
% , nograph % do not create graphs (which implies that they are not saved to the disk nor displayed) |
| 228 |
% , noprint % do not print stuff to console |
| 229 |
% , plot_priors = 1 % control plotting of priors |
| 230 |
% , prior_trunc = 1e-10 % probability of extreme values of the prior density that is ignored when computing bounds for the parameters |
| 231 |
% , TeX % print TeX tables and graphics |
| 232 |
|
| 233 |
% Data and model options |
| 234 |
% , first_obs = 501 % number of first observation |
| 235 |
% , logdata % if data is already in logs |
| 236 |
, nobs = 250 % number of observations |
| 237 |
% , prefilter=0 % demean each data series by its empirical mean and use centered moments |
| 238 |
|
| 239 |
% , xls_sheet = data % name/number of sheet with data in Excel |
| 240 |
% , xls_range = B2:D200 % range of data in Excel sheet |
| 241 |
|
| 242 |
% Optimization options that can be set by the user in the mod file, otherwise default values are provided |
| 243 |
% , huge_number=1e7 % value for replacing the infinite bounds on parameters by finite numbers. Used by some optimizers for numerical reasons |
| 244 |
@#ifdef NoEstim |
| 245 |
, mode_compute = 0 |
| 246 |
@#else |
| 247 |
, mode_compute = 13 % specifies the optimizer for minimization of moments distance |
| 248 |
, additional_optimizer_steps = [4] % vector of additional mode-finders run after mode_compute |
| 249 |
, mode_check % plot the target function for values around the computed minimum for each estimated parameter in turn |
| 250 |
@#endif |
| 251 |
% optim: a list of NAME and VALUE pairs to set options for the optimization routines. Available options depend on mode_compute, some exemplary common options: |
| 252 |
, optim = ('TolFun' , 1e-6 % termination tolerance on the function value, a positive scalar |
| 253 |
,'TolX' , 1e-6 % termination tolerance on x, a positive scalar |
| 254 |
,'MaxIter' , 3000 % maximum number of iterations allowed, a positive integer |
| 255 |
,'MaxFunEvals' , 1D6 % maximum number of function evaluations allowed, a positive integer |
| 256 |
% ,'UseParallel' , 1 % when true (and supported by optimizer) solver estimates gradients in parallel (using Matlab/Octave's parallel toolbox) |
| 257 |
% ,'Jacobian' , 'off' % when 'off' gradient-based solvers approximate Jacobian using finite differences; for GMM we can also pass the analytical Jacobian to gradient-based solvers by setting this 'on' |
| 258 |
) |
| 259 |
, silent_optimizer % run minimization of moments distance silently without displaying results or saving files in between |
| 260 |
|
| 261 |
% Numerical algorithms options |
| 262 |
% , aim_solver % Use AIM algorithm to compute perturbation approximation |
| 263 |
% , k_order_solver % use k_order_solver in higher order perturbation approximations |
| 264 |
% , dr=default % method used to compute the decision rule; possible values are DEFAULT, CYCLE_REDUCTION, LOGARITHMIC_REDUCTION |
| 265 |
% , dr_cycle_reduction_tol = 1e-7 % convergence criterion used in the cycle reduction algorithm |
| 266 |
% , dr_logarithmic_reduction_tol = 1e-12 % convergence criterion used in the logarithmic reduction algorithm |
| 267 |
% , dr_logarithmic_reduction_maxiter = 100 % maximum number of iterations used in the logarithmic reduction algorithm |
| 268 |
% , lyapunov = DEFAULT % algorithm used to solve lyapunov equations; possible values are DEFAULT, FIXED_POINT, DOUBLING, SQUARE_ROOT_SOLVER |
| 269 |
% , lyapunov_complex_threshold = 1e-15 % complex block threshold for the upper triangular matrix in symmetric Lyapunov equation solver |
| 270 |
% , lyapunov_fixed_point_tol = 1e-10 % convergence criterion used in the fixed point Lyapunov solver |
| 271 |
% , lyapunov_doubling_tol = 1e-16 % convergence criterion used in the doubling algorithm |
| 272 |
% , qz_criterium = 0.999999 % value used to split stable from unstable eigenvalues in reordering the Generalized Schur decomposition used for solving first order problems |
| 273 |
% , qz_zero_threshold = 1e-6 % value used to test if a generalized eigenvalue is 0/0 in the generalized Schur decomposition |
| 274 |
% , schur_vec_tol=1e-11 % tolerance level used to find nonstationary variables in Schur decomposition of the transition matrix |
| 275 |
% , mode_check_neighbourhood_size = 5 % width of the window (expressed in percentage deviation) around the computed minimum to be displayed on the diagnostic plots |
| 276 |
% , mode_check_symmetric_plots=1 % ensure that the check plots are symmetric around the minimum |
| 277 |
% , mode_check_number_of_points = 20 % number of points around the minimum where the target function is evaluated (for each parameter) |
| 278 |
); |