FORECAST OUT-OF-SAMPLE ....HELP!!!

[young econ]

Hi, I'm trying to construct an out of sample forecast for an output gap series.

I have estimated an AR(4) based on the latest 15 years of data (2000q1 to 2015q1).

y_gap c y_gap(-1 to -4)

I then want to forecast from 2015q2 to 2025q1. When I use the dynamic forecast the series remains constant, is there a way to construct a n-step ahead out of sample forecast that will have a variance that resembles the historical values?

Thanks in advance for your help.