Dear Professor Pfeifer,
I’ve got some problems on the policy design in the DSGE model. Would you please give any advice? Thank you very much.

I set a exogenous policy variable z_t in the model and its steady state is 1. I could easily get irf picture of some shock (e.g. technology shock) to the model when I just set z_t=1 (which means no policy improvement). But, If I want to get irf picture of some shock under the same time z_t is decreasing from 1 to 0, how could I realize this?

I read in some paper that the policy variable z_t is set as:
\log \left( \frac{{{z}_{t}}}{{\bar{z}}} \right)=\rho \log \left( \frac{{{z}_{t1}}}{{\bar{z}}} \right)\left( 1\rho \right)\phi \log \left( \frac{{{y}_{t}}}{{\bar{y}}} \right) in which y represents the output. The value of parameter \rho means the persistence of the policy. How is this persistence related to the real time? For example, if \rho=0.5, how many quarters is the persistence of the policy? 
If I want to set \log \left( \frac{{{z}_{t}}}{{\bar{z}}} \right)=\rho \log \left( \frac{{{z}_{t1}}}{{\bar{z}}} \right)\left( 1\rho \right)\phi \log \left( \frac{{{y}_{t}}}{{\bar{y}}} \right) when output is positively deviate its steady state( \frac{{{y}_{t}}}{{\bar{y}}}>1) and z_t=1 when output is negatively deviate its steady state( \frac{{{y}_{t}}}{{\bar{y}}}<1), how should I realize it?
Thanks again for your time on this.