Data handling

I have a question about handling observed series when estimating a model with dynare. Suppose I write my model in nonlinear terms, i.e. variables are expressed in levels, not in deviations from their steady state values. Also, suppose there are no trends in the exogenous variables processes. Then, should I de-mean the observed data before estimation or not? My sense is that, for consistency, the observed variable should not be de-meaned if the corresponding model variable is in levels, but I’m not quite sure.
Can you help me on this?

Sorry for the delay in answering. You are right, you shouldn’t demean, but make sure that you are estimating enough parameters. Basically, you need to make sure that if you were to change the units of the observed variables, that you have in the model parameters that could change value to accomodate that change of units.