Page 1 of 1

A question about how to variance decomposition!

PostPosted: Sun May 26, 2013 7:01 am
by yoyoNDHU
Hello~~Sorry I have a question again..

I want to ask how to get the posterior variance decomposition in percentage~

here is my code:
estimation(datafile=datayo2,nobs=100,first_obs=10,mh_replic=2000,mh_nblocks=2,mh_drop=0.45,mh_jscale= 0.8,mode_compute=4,bayesian_irf,forecast=10,moments_varendo)y ch I ce ke omn;

but the result in oo_.PosteriorTheoretical.dsge.variance decomposition is not percentage~
Should I need to do the percentage by myself ?

or I should use

estimation(datafile=datayo2,nobs=100,first_obs=10,mh_replic=2000,mh_nblocks=2,mh_drop=0.45,mh_jscale= 0.8,mode_compute=4,bayesian_irf,forecast=10,moments_varendo,)y ch I ce ke omn;
stoch_simul(period=2100, conditional_variance_decomposition=1)

:?: ~~Thanks for everyone's help!~~

Best regards YOYO

Re: A question about how to variance decomposition!

PostPosted: Sun May 26, 2013 7:14 am
by jpfeifer
What do you mean with "It's not in percentage"?

The first option will give you the mean/median variance decomposition while the second one will give you the decomposition at the mean/median. This is an important difference in terms of interpretation.

Re: A question about how to variance decomposition!

PostPosted: Sun May 26, 2013 7:49 am
by yoyoNDHU
Thank you jpfeifer~!

I means for example the result from "moments_varendo"

I got consumption like :
ea =0.209
ey=0.321
epi=0
er=0
eu=0.005
ec=0.003
el=0.306
emo=0.164
em=0.005
euh=0.007
eo=0.323

but the sum of these will not equal one(100%)~~Should I need to do the percentage by myself??

Thank you!

Re: A question about how to variance decomposition!

PostPosted: Sun May 26, 2013 4:01 pm
by yoyoNDHU
Dear jpfeifer~

Here my mod file, when I use

stoch_simul(order=1,irf=0,conditional_variance_decomposition=[1,4]);

it's OK~I can get the variance decomposition and conditional variance decomposition (in percent)

but in the estimate side

I use

estimation(datafile=ddyo2,nobs=40,first_obs=50,mh_replic=2000,mh_nblocks=2,mh_drop=0.45,mh_jscale= 0.8,mode_compute=4,bayesian_irf, moments_varendo,conditional_variance_decomposition=1)y ch I o eps q ce ke omn;

but I get error...,but if I don't put "conditional_variance_decomposition=1" in the estimation

I don't get error but the variance decomposition in oo_.Posterior......isn't in percent.

Sorry , my English is really poor ..., but I really need your help ~
Thank you

Best Regards YOYO

{??? Subscripted assignment dimension mismatch.

Error in ==> <a href="matlab: opentoline('D:\4.3.2\matlab\conditional_variance_decomposition_mc_analysis.m',66,0)">conditional_variance_decomposition_mc_analysis at 66</a>
tmp(i1:i2,:) =
transpose(dynare_squeeze(Conditional_decomposition_array(endogenous_variable_index,:,exogenous_variable_index,:)));

Error in ==> <a href="matlab: opentoline('D:\4.3.2\matlab\posterior_analysis.m',78,0)">posterior_analysis>job at 78</a>
oo_ =
conditional_variance_decomposition_mc_analysis(SampleSize,'posterior',M_.dname,M_.fname,..