Zero steady state values and percentage deviations

  1. Log-linearization typically means also subtracting the log mean.
  2. You hardly ever need a transformation of the model variables to make them “stationary”. Transformations are only made to make the model variables comparable to the data.