One variable has 0 steady state value, the model must be linearized?

dear professor Pfeifer;
I have been reading your paper “Macroeconomic Effects of Financial Shocks: Comment”, the steady state value of chi {\chi} (long_name=‘Lagrange Multiplier demand constraint’) is 0, it seems that you write the orginal model and use linear model estimation. I want to know if I want to write down linear model,do I have to linearize the whole system, or can I loglinearize the whole system except chi, since this is much easier and more intuitive.


It always depends on what you want to do. My reply at Question about understanding irfs in dynare should be useful to understand different approaches to linearization an log-linearization. The short answer to your question is: you can work on a variable by variable basis. Variables already in percent or with steady state 0 can be linearized while everything else is log-linearized.

thanks professor Pfeifer, now I get it~