Estimation/making observed variables comparable with model

I am about to estimate a variant of Kollmann (2001, 2002) extended with a variety of shocks and rigidities. I have done so before using Chris Sims’ GENSYS, but this time I should like to use DYNARE.

The model is without trends and stated in levels, (meaning I have not have not linearized it).

Having read the DYNARE documentation I still seek answers/advice to the following two questions:

  1. Which estimation option is preferable in my situation? Linear or loglinear?
  2. How should I transform my data (GDP for example) in order make sure that the unit of measurement of the variables in the model is comparable to those of the observed variables?

Jesper Linaa

  1. As you are modeling a growing economy, it would be better to take growth into account in your model

  2. As your model, for now doesn’t take growth into account, you need to remove the trend from the growing variables. Doing that, you most likely also remove the mean. The observed variables in your model should be in deviation from the mean

  3. Thea advantage of loglinearizing in that your observed variables are unit free percentage deviation from the mean, so you don’t have to worry about what are the units of your detrended variables.

  4. I would write the model in log, rather than relying on the loglinear option of estimation. This will give you a better control on what you are doing.