Decay Process of AR1 Forecasts

[miksterz]

Hello,

I need to generate 12-step ahead out of sample forecasts for a basic AR(1) model

By construction the forecasts generated from an AR(1) process have to display a decaying property. However, I do not seem to be getting this result. Instead my forecast series is upward sloping and then sort of levels off

I am using least squares to estimate the equation from 1986-2000

Then I am using the command "EQ01.forecast" for a forecast sample of 2001-2010

Can somebody tell me what I might be doing wrong?

Thanks