Just want to verify some details about the model solving in Dynare.
What kind of method is used to solve the linearized model in dynare? (Sim’s QZ decompostion, right? ) 2.It seems that the result can be expressed in the form
Y(t)=AY(t-1)+Be(t)
and all the variables are put in Y. Then, measurement equation is part of the above equation, right? Is there a straightford way to extract the measurement equations?
Right, but this is more like a transition equation
When Dynare is used for estimation, the measurement equation
has the following form
[code]
Y*(t) = MYbar + MY(t)+X(t) + eta(t)
[code]
where Y*(t) are the observed variables
M is a selection matrix (one only observes a few variables in general)
Ybar is the steady state (this assumes Y(t) represents deviation from steady state)
X(t) is an exogenous linear trend (using observation_trend)
eta(t) is a vector of measurement errors