Consider a stationary model that combines the \(\operatorname{AR}(2)\) model \(y_{t}=\delta+\theta_{1} y_{t-1}+\theta_{2} y_{t-2}+e_{t}\) with an \(\mathrm{AR}(1)\) error model \(e_{t}=ho e_{t-1}+v_{t}\) where \(E\left(v_{t} \mid I_{t-1}\right)=0\). Show that

Why will the assumption \(E\left(y_{t} \mid I_{t-1}\right)=\delta+\theta_{1} y_{t-1}+\theta_{2} y_{t-2}\) be violated if the errors are autocorrelated?

Transcribed Image Text:

E(y-1)=8(1p)+(0, +p) y +(02-01p)y-2-02py-3