No, this is not a proper way of loglinearization. It seems you are trying to do an exp()-substitution, but in this case, the steady state values are not 0. You may want to consult Linearisation vs Log-linearisation and Question about understanding irfs in dynare for your options.