Dear Dynare Team,
with the publication “Guide for specifying…” and the forum posts I was capable of transforming the data to match the variables of a loglinearized model by removing the trend via log difference and demeaning.
However, I would like to gain a better understanding of what I’m actually doing instead of solely replicating the necessary steps.
I hope, one of you can answer me the following questions:
 The observation equation for output is dy=yy(1). This makes sense to me since we have to match the GDP growth rate dy with the model variables (that is (log(y)log(y(1)).
However, I don’t understand why the observation equation for inflation is infobs=inf.
For inflation we also apply the log difference and demean it right? So we would also get the growth rate of Inflation. Why isn’t it infobs=infinf(1) ? What is the difference between Output and Inflation?

Is it correct that the interest rate is not trending and thus we only apply logs instead of log differences and demean it?

If we want to estimate Bayesian irfs, do we usually estimate y,inf,i or is it common to estimate dy,infobs,iobs – again: what is the difference ?
I’ve read the Remark of Figure 10 in “An Introduction to Graphs in Dynare” but I don’t understand what output do we get for each of the two types
 Is the output of bayesian estimation (Graphs for Bayesian Irfs) given in basis points? E.g. inflation decreased by 30 basis points below steady state ?
Thank you very much for help and clarification!
Johanna