Dear all.

I try to compare the f models using by DSGE-prior-weight lambda.

In order to do this, I want to compute the relative indicator proposed by Adolfson et al.(2008). This indicator is defined as the deviation of optimal lambda from minimal lambda.

Where can I find this minimal lambda in results of estimation?

(Of course, I know that dynare compute this minimal value automatically since some error message shows me this value.)

Best regards.

Hi, I do not know Adolfson et al. (2008), so I am not sure I understand what you are looking for. If the minimal \lambda is the minimal value of the DSGE prior weight consistent with a proper inverted Wishart – Gaussian prior. This value is:

\underline{\lambda} = \frac{mp+m}{T}

where m is the number of variables in the VAR model, p the number of lags and T the number of periods in the sample. I don’t think we save this under `oo_`

, but it’s so simple that it would not be really useful. Note that this threshold is independent of the underlying DSGE model. So if you only change the DSGE prior and not the VAR model, you can directly compare the unnormalized \lambda. More details on the prior is provided on the wiki:

Best,

Stéphane.

Thanks! That is exactly I want to know.

I would like to compare the lambda simply.

Sincerely yours.