Dear Johannes, Thank you very much for your answer, I will keep that in mind.

What I am trying to do is to incorporate average inflation targeting in Gambacorta and Signoretti (NK_GS14 in mmb) in order to run stochastic simulations and compare them to those under standard inflation targeting. Then I would like to change same parameters linked to the financial system as the level of indebtedness in the economy and see how the outcome changes given different levels of leverage in the economy.

As the model is non linear my current average inflation targeting rule for a 2 years window is the following:

```
1) exp(plev)=exp(plev(-1)+pie); %price level
2) exp(paitwo)=exp(plev-plev(-8)); %AIT2
3) (1+r_ib) = (1+r_ib(-1))^rho_ib*(1+r_ib_ss)^(1-rho_ib)*(((exp(paitwo)/(piss)))^phi_pie*((exp(Y1))/(exp(steady_state(Y1))))^phi_y)^(1-rho_ib) ; %TR
```

Now the problem that I have is that using this approach dynare is not able to find the steady state. If I instead define `exp(paitwo)`

as `exp(pie+pie(-1)+pie(-2)+pie(-3)+pie(-4)+pie(-5)+pie(-6)+pie(-7))`

which should be equal to `exp(plev-plev(-8))`

given equation 1 dynare is actually able to find it but the irfs look pretty weird and oscillating.

Now, I am trying also a different approach where average quarterly inflation over 2 years is defined as exp(paitwo)=exp((pie+pie(-1)+pie(-2)+pie(-3)+pie(-4)+pie(-5)+pie(-6)+pie(-7))/8) actually I obtain much better outcomes.

In the steady state there is no inflation and the initial value for pie is 0.

What do yo think is the source of problems here? Maybe time convention? Some insights would be really appreciated, thank you in advance.