path for endogenous variables

[marzie]

Hi,

I have difficulty computing path of endogenous variables under *given* path of exogenous shocks.
given the instruction in Dynare manual : http://www.dynare.org/manual/index_23.html , I use this notation: Y(t)= Y_s + A * (Y(t-1) -Y_s) + B* u(t)

The main difficulty is to compute the second term in the above model as the rank of A is not compatible with (Y(t-1)-Y(s)). This is because A is the policy transition matrix which links the state variables to the endogenous variables , so it is not a square matrix. Normally I augment A manually to make it a square matrix but this is very time consuming. I wonder if there is any short cut?

In a way what I want to do is similar to the 'extended path' function which produces ' oo_.endo_simul ' but with only one differece that I already have the path for the exogenous variables. I looked at the source code of 'extended path' http://www.dynare.org/dynare-matlab-m2h ... _path.html , I am not sure if I can simply modify this code by giving my U(t) as 'exo_simul'. Not sure if it is the righ way to do it.


Thank you
Marzie

As it is explained in the manual: the coefficients of the decision rules are stored as follows:

is stored in oo_.dr.ys. The vector rows correspond to all endogenous in the declaration order.
A is stored in oo_.dr.ghx. The matrix rows correspond to all endogenous in DR-order. The matrix columns correspond to state variables in DR-order.
B is stored oo_.dr.ghu. The matrix rows correspond to all endogenous in DR-order. The matrix columns correspond to exogenous variables in declaration order.

--------------------------------------------------------------------------------

[SébastienVillemot]

Hi,

I have difficulty computing path of endogenous variables under *given* path of exogenous shocks.
given the instruction in Dynare manual : http://www.dynare.org/manual/index_23.html , I use this notation: Y(t)= Y_s + A * (Y(t-1) -Y_s) + B* u(t)

The main difficulty is to compute the second term in the above model as the rank of A is not compatible with (Y(t-1)-Y(s)). This is because A is the policy transition matrix which links the state variables to the endogenous variables , so it is not a square matrix. Normally I augment A manually to make it a square matrix but this is very time consuming. I wonder if there is any short cut?

There is a shortcut: the simult_.m function provided by Dynare. It is not well documented, but it is easy to use.