Page 1 of 1
maximum likelihood estimation
Posted:
Sat Mar 28, 2009 11:21 amby alexnomad
Hello all,
I want to estimate some model by maximum likelihood but some of the variables are exogeneous and I don't want to build an equation for exogeneous variables. Is it possible to estimate by ML without specifying the DGP of exogeneouv variables ?
Thanks in advance
Re: maximum likelihood estimation
Posted:
Sat Mar 28, 2009 1:00 pmby StephaneAdjemian
This is not possible with Dynare. The likelihood is the density of the sample conditional on the model and its parameters. The likelihood is built from the DGP, consequently we cannot estimate a model without specifying the model for the exogenous variables. Is your exogenous variable latent or observed?
Best, Stéphane.
Re: maximum likelihood estimation
Posted:
Sat Mar 28, 2009 6:31 pmby alexnomad
thanks Stephane.
My exogeneous variables are observed
Re: maximum likelihood estimation
Posted:
Mon Mar 30, 2009 9:02 amby StephaneAdjemian
If, the exogeneous variable is, for instance, the price of oil, and if you don't want to write structural equations for this variable, you can still assume an ad hoc autoregressive process (or anything else) for this variable.
Stéphane.