I have a question regarding the DSGE VAR option in estimation: Does this trigger the estimation of a DSGE VAR or a DSGE VECM (as in Del Negro/ Schorfheide/ Smets Wouters (2004), attached)? My take from Del Negro/ Schorfheide/ Smets Wouters (2004) was that a VAR in growth rates without error correction terms approximates the DSGE IRFs only for a very large number of lags, while a VECM requires only a normal number of lags to do so. Many thanks,
Sorry, this here is the paper.
DelNegroSchorfheideSmetsWouters.pdf (347.3 KB)
Dynare’s implementation is the Del Negro/Schorfheide (2004) one, i.e. there is no error correction term. See also http://www.dynare.org/DynareWiki/DsgeVar