Thanks Johannes, see below

On 09/07/2012 09:19 AM, Johannes Pfeifer wrote:

Dear all,

I have two issues with conditional_variance_decomposition.

1. The option conditional_variance_decomposition in stoch_simul only triggers the decomposition, if theoretical moments are requested, i.e. periods=0. I guess this makes sense as the decomposition is based on the state-space form, which is a theoretical concept. However, we never provide a warning that no variance decomposition will be performed nor do we document the requirement in the manual.

We should add a warning in the code and in the manual

2. Even if one specifies order=2 (with periods=0), Dynare still outputs a variance_decomposition. If I understand it correctly, I am not sure this is a good idea as the decomposition is based on the first order state-space system. We should either shut off this feature for order=2 or at least provide an explicit warning that the decomposition is only for first order although a higher order approximation was used.

We should add a warning for all second order moments as the same is true for variance, correlations and autocorrelations

Please, enter a trac ticket

Best

Michel

Regards,

Johannes

--------

Johannes Pfeifer

Haußerstr. 29

72076 Tübingen

Tel.: +49-(0)7071-6396184

Mobil: +49-(0)170-6936820

jpfeifer@gmx.de
_______________________________________________
Dev mailing list
Dev@dynare.org
https://www.dynare.org/cgi-bin/mailman/listinfo/dev