Page 1 of 1

question on conditional variance series

PostPosted: Thu Aug 15, 2013 9:31 pm
by YiJohn
Hello everyone/Dr. jpfeifer,

I was solving a standard DSGE model, and I was looking for conditional variance of endogenous variables, like var(y_{t+k}|t) with respect to different horizon k.

As Dr. jpfeifer suggested in http://www.dynare.org/phpBB3/viewtopic.php?f=1&t=4844, I changed conditional_variance_decomposition.m to get the conditional variance series.

Code: Select all
save conditionalvariance ConditionalVariance;


My aim is comparative static analysis: the effect of different parameter values on conditional variance series.

My problem: no matter how I adjust the parameter values (including elasticity of substitution, and the exogenous variances), the conditional variance series never change, so I was wondering this variable "ConditionalVariance" in conditional_variance_decomposition.m, is it already treated/normalized by dynare?

Thanks for your help in advance!

Yi

Re: question on conditional variance series

PostPosted: Mon Aug 19, 2013 8:42 am
by jpfeifer
Could you please post the mod-file. I cannot replicate the described behavior.

Re: question on conditional variance series

PostPosted: Mon Aug 26, 2013 6:39 am
by YiJohn
Hi Dr. jpfeifer,

Sorry for my belated reply.

Please see attached my mod-file and the modified matlab file. The problem is: No matter how I adjust the parameters "tau"(0<tau<1) or "phi"(0<phi<1), the conditional variance series stay the same.

I am aware that the ASYMPTOTIC variances should be independent of "tau" and "phi", but I believe CONDITIONAL variance series should be sensitive to "tau" and "phi".

Thank you very much for your help!

Best Regards,
Yi
revision.mod
(1.71 KiB) Downloaded 76 times

conditional_variance_decomposition.m
(3.3 KiB) Downloaded 71 times

Re: question on conditional variance series

PostPosted: Mon Aug 26, 2013 9:06 am
by jpfeifer
Most other parameters change the conditional variance. Thus, it seems to be an economic feature of your model.

Re: question on conditional variance series

PostPosted: Tue Aug 27, 2013 3:16 am
by YiJohn
Dr. jpfeifer,

Thanks a lot!

Regards, Yi

Re: question on conditional variance series

PostPosted: Tue Aug 27, 2013 6:39 am
by YiJohn
Hello Dr. jpfeifer,

Could you please tell me which parameter values you tried? Thanks in advance. Because as I tried different parameter values, the conditional variance series do not change.

The following is how I calculate the AGGREGATE conditional variance series (AGGREGATE means subject to all the shocks). Please let me know if I made any mistakes. Thanks a lot.

Since I was trying to get the aggregate conditional variance series, after I run "dynare revision", I use the following matlab code to get the aggregate conditional variance series:
Code: Select all
load conditionalvariance
A = sum(ConditionalVariance,3);
B = A';


Then the rows of matrix B is for different periods, and the columns of matrix B is for different variables.

Thanks again for your kind help!

Regards,
Yi

Re: question on conditional variance series

PostPosted: Tue Aug 27, 2013 2:40 pm
by jpfeifer
For example, I changed the variance in the shocks-block and beta.

Re: question on conditional variance series

PostPosted: Tue Aug 27, 2013 9:31 pm
by YiJohn
Dr. jpfeifer,

Thanks a lot for your reply.

Best,
Yi