stepan
/
dynare
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Make examples/fs2000.mod closer to the original code. 

- Use (old default) mode_compute=4 which is closer to the algorithm
   used by Frank Schorfheide and ensures that the hessian matrix is well
   behaved (contrary to the new default, because of the asymptote at 0
   in the beta prior for autoregressive parameter ρ).

 - Change parameterization for mst. A normal prior on mst is not
   equivalent to a normal prior on log(mst) (which is done the
   parameterization in the JAE paper).

Closes #2177.
mr#2177
Stéphane Adjemian (Guts) 2023-12-14 12:09:09 +01:00
parent 162813225d
commit 1983dc13a3
Signed by: stepan
GPG Key ID: 295C1FE89E17EB3C
2 changed files with 29 additions and 29 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

50
examples/fs2000.mod
View File

@ -1,17 +1,17 @@
/*
* This file replicates the estimation of the cash in advance model (termed M1
* in the paper) described in Frank Schorfheide (2000): "Loss function-based
* This file replicates the estimation of the cash in advance model (termed M1
* in the paper) described in Frank Schorfheide (2000): "Loss function-based
* evaluation of DSGE models", Journal of Applied Econometrics, 15(6), 645-670.
*
* The data are taken from the replication package at
* The data are taken from the replication package at
* http://dx.doi.org/10.15456/jae.2022314.0708799949
*
* The prior distribution follows the one originally specified in Schorfheide's
* paper. Note that the elicited beta prior for rho in the paper
* implies an asymptote and corresponding prior mode at 0. It is generally
* recommended to avoid this extreme type of prior.
* recommended to avoid this extreme type of prior.
*
* Because the data are already logged and we use the loglinear option to conduct
* Because the data are already logged and we use the loglinear option to conduct
* a full log-linearization, we need to use the logdata option.
*
* The equations are taken from J. Nason and T. Cogley (1994): "Testing the
@ -62,20 +62,20 @@ varexo e_a ${\epsilon_A}$ (long_name='TFP shock')
e_m ${\epsilon_M}$ (long_name='Money growth shock')
;
parameters alp ${\alpha}$ (long_name='capital share')
bet ${\beta}$ (long_name='discount factor')
gam ${\gamma}$ (long_name='long-run TFP growth')
mst ${m^*}$ (long_name='long-run money growth')
rho ${\rho}$ (long_name='autocorrelation money growth')
phi ${\phi}$ (long_name='labor weight in consumption')
del ${\delta}$ (long_name='depreciation rate')
parameters alp ${\alpha}$ (long_name='capital share')
bet ${\beta}$ (long_name='discount factor')
gam ${\gamma}$ (long_name='long-run TFP growth')
logmst ${\log(m^*)}$ (long_name='long-run money growth')
rho ${\rho}$ (long_name='autocorrelation money growth')
phi ${\phi}$ (long_name='labor weight in consumption')
del ${\delta}$ (long_name='depreciation rate')
;
% roughly picked values to allow simulating the model before estimation
alp = 0.33;
bet = 0.99;
gam = 0.003;
mst = 1.011;
logmst = log(1.011);
rho = 0.7;
phi = 0.787;
del = 0.02;
@ -84,7 +84,7 @@ model;
[name='NC before eq. (1), TFP growth equation']
dA = exp(gam+e_a);
[name='NC eq. (2), money growth rate']
log(m) = (1-rho)*log(mst) + rho*log(m(-1))+e_m;
log(m) = (1-rho)*logmst + rho*log(m(-1))+e_m;
[name='NC eq. (A1), Euler equation']
-P/(c(+1)*P(+1)*m)+bet*P(+1)*(alp*exp(-alp*(gam+log(e(+1))))*k^(alp-1)*n(+1)^(1-alp)+(1-del)*exp(-(gam+log(e(+1)))))/(c(+2)*P(+2)*m(+1))=0;
[name='NC below eq. (A1), firm borrowing constraint']
@ -119,25 +119,25 @@ end;
steady_state_model;
dA = exp(gam);
gst = 1/dA;
m = mst;
m = exp(logmst);
khst = ( (1-gst*bet*(1-del)) / (alp*gst^alp*bet) )^(1/(alp-1));
xist = ( ((khst*gst)^alp - (1-gst*(1-del))*khst)/mst )^(-1);
nust = phi*mst^2/( (1-alp)*(1-phi)*bet*gst^alp*khst^alp );
xist = ( ((khst*gst)^alp - (1-gst*(1-del))*khst)/m )^(-1);
nust = phi*m^2/( (1-alp)*(1-phi)*bet*gst^alp*khst^alp );
n = xist/(nust+xist);
P = xist + nust;
k = khst*n;
l = phi*mst*n/( (1-phi)*(1-n) );
c = mst/P;
d = l - mst + 1;
l = phi*m*n/( (1-phi)*(1-n) );
c = m/P;
d = l - m + 1;
y = k^alp*n^(1-alp)*gst^alp;
R = mst/bet;
R = m/bet;
W = l/n;
ist = y-c;
q = 1 - d;
e = 1;
gp_obs = m/dA;
gy_obs = dA;
end;
@ -151,7 +151,7 @@ estimated_params;
alp, beta_pdf, 0.356, 0.02;
bet, beta_pdf, 0.993, 0.002;
gam, normal_pdf, 0.0085, 0.003;
mst, normal_pdf, 1.0002, 0.007;
logmst, normal_pdf, 0.0002, 0.007;
rho, beta_pdf, 0.129, 0.223;
phi, beta_pdf, 0.65, 0.05;
del, beta_pdf, 0.01, 0.005;
@ -161,8 +161,8 @@ end;
varobs gp_obs gy_obs;
estimation(order=1, datafile=fs2000_data, loglinear,logdata, mh_replic=2000, mh_nblocks=2, mh_jscale=0.8, mode_check);
estimation(order=1, datafile=fs2000_data, loglinear,logdata, mode_compute=4, mh_replic=20000, nodiagnostic, mh_nblocks=2, mh_jscale=0.8, mode_check);
%uncomment the following lines to generate LaTeX-code of the model equations
%write_latex_original_model(write_equation_tags);
%collect_latex_files;
%collect_latex_files;

8
examples/fs2000_data.m
View File

@ -1,6 +1,6 @@
%This file is a direct Matlab implementation of the loaddata.g and data.prn files
%of Schorfheide, Frank (2000): Loss function-based evaluation of DSGE models
%(replication data). Version: 1. Journal of Applied Econometrics. Dataset.
%of Schorfheide, Frank (2000): Loss function-based evaluation of DSGE models
%(replication data). Version: 1. Journal of Applied Econometrics. Dataset.
%http://dx.doi.org/10.15456/jae.2022314.0708799949
% Copyright: 2000-2022 Frank Schorfheide
% Copyright: 2023 Dynare Team
@ -210,6 +210,6 @@ data_q=[18.02 1474.5 150.2
%Compute growth rates: from 1950:I to 1997:IV
gy_obs=1000*data_q(:,2)./data_q(:,3); %real GDP per capita
gy_obs=diff(log(gy_obs));
gy_obs=diff(log(gy_obs));
gp_obs = diff(log(data_q(:,1))); %GDP deflator inflation
gp_obs = diff(log(data_q(:,1))); %GDP deflator inflation