Log-lin & lin together

[dyndyn]

Is it possible to put a non-linear model into Dynare and ask Dynare to log-linearize some variables and to linearize some others?
I would put the variables (X) that I want in log-lin in the form "exp(X)" while I would put the other ones without exp.
Can those kinds of variables stay together in a model?
For example, I want the deviation of the interest rate, not the log-deviation.

Thanks

[jpfeifer]

See Remark 19 (Variables with Negative Steady States or already in Percent) in Pfeifer(2013): "A Guide to Specifying Observation Equations for the Estimation of DSGE Models" https://sites.google.com/site/pfeiferecon/Pfeifer_2013_Observation_Equations.pdf.

[dyndyn]

Thank you. I got the point. Just another question. Does Remark 19 mean that it is "better" to put GROSS interest or inflation rates in log rather than in levels?

[jpfeifer]

If they are close to 1, it does not really matter. In that case, percentage and absolute deviations are the same as the base is approximately 1. Moreover, what is the metric for "better"? It all depends on the interpretation you want
.

[dyndyn]

I want to see the effect of a change in r (from, say, 3% to 4%) and in my model I have the variable R=(1+r). Should it be R or exp(R)? Thanks

[jpfeifer]

That should not make any difference. Regardless of variable transformations, the equation must be the same, only the variable interpretation differs. Please be more specific and post the particular exogenous process/equation you are talking about.