Exogenous variables in state space
Posted: Wed Dec 16, 2009 1:32 pmHi!
I want to estimate a model, in which I have observations on exogenous variables, that possibly
enter the transition equations in state space form. As far as I know Dynare does not allow observations for
exogenous variables.
I tried to solve this problem by defining my otherwise exogenous variables as endogenous and then create a
"dummy" exogenous variable for each one of them. ( Eg.: endo_var was an observed exogenous variable in one of the equations, then I create an exo_var exogenous variable and add the eq endo_var = exo_var.) The only set back of this approach is that the variance of my observations will enter the likelihood calculations this way (regarded as shock variance!), making max likelihood or bayesian estimation erroneous ?! (If I'm right, that is)
Anyone have a good comment on this, please ?
I want to estimate a model, in which I have observations on exogenous variables, that possibly
enter the transition equations in state space form. As far as I know Dynare does not allow observations for
exogenous variables.
I tried to solve this problem by defining my otherwise exogenous variables as endogenous and then create a
"dummy" exogenous variable for each one of them. ( Eg.: endo_var was an observed exogenous variable in one of the equations, then I create an exo_var exogenous variable and add the eq endo_var = exo_var.) The only set back of this approach is that the variance of my observations will enter the likelihood calculations this way (regarded as shock variance!), making max likelihood or bayesian estimation erroneous ?! (If I'm right, that is)
Anyone have a good comment on this, please ?