I am using a slightly readapted model of Gerali et al (2010) and the model is non-linear and all the vars are expressed as exp(.). This is a functioning code file I got from someone else, but my questions is, when I want to use different parameters to apply the model for a different country, and I am mainly interested in producing IRFs, do I need to change the initial values (they are not analytical, but numerical initial values) assuming the code still runs?
So if I use the original initial values that were for a country A, and I change the parameters to fit the characteristics of the country B, and A and B are both small open economies but certainly different in some aspects, would using the original initial values produced biased results?