Page 1 of 1

Forecast using bayesian estimation

PostPosted: Tue May 29, 2007 1:20 am
by wgonzalez
I am trying to reply the Smets and Wouter (2003) work , in which I want to make forecast for the euro zone but I have an error when trying to carry out it, I would thank them enough if could tell me how can I solve that problem. The error finishing the algorithm metropolis - hastings is

?? Error using ==> plus
Matrix dimensions must agree.

Error in ==> <a href="error:C:\dynare_v3\matlab\metropolis.m,2039,1">metropolis at 2039</a>
yf(:,IdObs) = yf(:,IdObs)+repmat(bayestopt_.mean_varobs', ...

Error in ==> <a href="error:C:\dynare_v3\matlab\dynare_estimation.m,778,1">dynare_estimation at 778</a>
metropolis(xparam1,invhess,gend,data,rawdata,bounds);

Error in ==> <a href="error:C:\Documents and Settings\Administrador\Escritorio\sw\sw6.m,277,1">sw6 at 277</a>
dynare_estimation(var_list_);

Error in ==> <a href="error:C:\dynare_v3\matlab\dynare.m,26,1">dynare at 26</a>
evalin('base',fname) ;


thanks in advance

WGonzalez

PostPosted: Tue Apr 22, 2008 8:27 pm
by jtsoukalas
Dear WGonzalez

I think you are estimating with prefilter option == 1.


Simply remove the transpose from bayestopt_.mean_varobs' in this and several lines down in the metropolis.m file. I think it should work.

JT

Error in ==> <a href="error:C:\dynare_v3\matlab\metropolis.m,2039,1">metropolis at 2039</a>
yf(:,IdObs) = yf(:,IdObs)+repmat(bayestopt_.mean_varobs'