Little doubt on estimation

Hi guys, I’m in the middle of a struggle with my thesis, and a small disturbance crossed on my way. Perhaps one of you that are doing, or did the same as I do at this time could lend my a hand.
The thing is that, I’m estimating a simple NK model, with three observed variables x, pi and r (in deviations with respect from the output gap, and from the steady state the two latest) . As usual, I employ the log of the GDP (serie), and then I use the cyclical componen after hp filter it. But my question is, should I detrend the two other variables? In theory are also in deviations so I should, but I’m somehow hesitant…
I’d be pleased if anyone show me the light!

It depends on how you map your data to your model variables, i.e. how you specify your observation equations. Please search the forum for “observation equation”. There are many posts dealing with this problem.

Thanks buddy!