Estimation with random walk

Hello all.
I am trying to estimate by maximum likelihood model which includes random walk equation. When optimization starts dynare stack on the same value function in iterations and shows NaN in each iteration. How effectively I can estimate this model ? Thanks a lot in advance

There is a detailed example in the Dynare user guide (chapter six).


There is some issue when one uses Dynare 4 to estimate a model with unit roots + deterministic trends + measurement error. One workaround is to treat measurement errors as additional shocks to the observation equations inside the model block. I posted a message a while ago but noone answered :cry: