Is unit_root_vars obsolete?

I have read in the manual that,

“unit_root_vars is now deprecated and will result in no action, It was used to declare unit-root variables of a model so that a diffuse prior can be used in the initialization of the Kalman filter for these variables only.”

I am not sure exactly what is meant by the word “deprecated”.

Do I need to use this argument when my observed variables are random walks plus a drift? When I use the argument, Dynare seems to run and work properly. Without the argument, Dynare crashes.

Could some one please give me a status update on this argument. Is it needed? Can I do without this argument if I re-write my _steadystate.m file (by taking out the Y_obs = 1; lines for the trending observations perhaps?)

Thank you.

You shouldn’t need to use unit_root_variables. What is the error message when Dynare is crashing?



When I use unit_root_vars in the mod file I do not get an error and the estimation command completes successfully. Without unit_root_vars in the mod file I obtain the following:

??? Error using ==> initial_estimation_checks at 92
The seadystate values returned by BY_closed_extra_shocks_rwtech_steadystate.m don’t solve the static model!

Error in ==> dynare_estimation_1 at 334

Error in ==> dynare_estimation at 62

Error in ==> BY_closed_extra_shocks_rwtech at 218

Error in ==> dynare at 132

This error occurs immediately after the data is read.
evalin(‘base’,fname) ;

My variables are C_obs, Y_obs, and W_obs. All are trending, so I had the following line in my mod file:

unit_root_vars C_obs, Y_obs W_obs;

and I also set the steady state value of these variables in my steady_state.m file to 1.

If I remove the unit_root_vars line in the mod file, should I remove the steady state values C_obs=1, Y_obs=1, and W_obs=1 in my steady_state.m file? Or give them some other value?