Degree of freedom in DSGE-VAR

[earsmall]

Hi, my question is not directly related to Dynare programming.
It is about the theory in DSGE-VAR with reference to http://www.dynare.org/DynareWiki/DsgeVar.

The question is, the prior distribution of forecasting error variance in DSGE-VAR follows an inverted Wishart distribution
with degree of freedom LT-mp-m where LT is the number of artificial observations generated from DSGE model, m is the number of
VAR model variables and p is the lag length.

Is there anyone who can explain why the degree of freedom is LT-mp-m ?
or Could you introduce me some internet pages including proper explanation?

Thank you.

[jpfeifer]

Please take a look at Del Negro/Schorfheide (2004): Priors from general equilibrium models for VARs, International Economic Review. We have to check the page you refer to. The treatment of the constant term seems strange.