Weird behavior Taylor Rule

Dear all,
I am facing a weird issue with Dynare 4.3.3. Basically using the Taylor Rule

r_t = r_ss*(infl/inflSS)^(1.5)*(GDP/GDP(-1))^(1.5)

what happens is that when GDP actually increases, the interest rates goes down (and viceversa). Setting the coefficient for GDP to 0, r follows inflation as it should.
Has anyone ever faced the same problem or have ideas about how to solve it?

Thank you

Structural shocks that affect the economy will always affect both inflation and output. Your interpretation relies on an intuition where you exogenously vary GDP, keeping everything constant. But this does not happen as inflation will move and this equilibrium behavior will take the Taylor rule into account. Take a positive TFP shock. It will increase output and decrease inflation. Depending on the strength of the responses and the respective weights in the Taylor rule, the nominal interest rate might decrease and thus negatively comove with output.

This is also the reason why, after a positive monetary policy shock, it can happen that the nominal interest rate decreases.

Dear Johannes,
thanks for you reply. I see what you mean, especially in case of sticky prices for which GDP and inflation are strongly related.