= Periodic Yule-Walker equations
For unit-variance innovations, orthogonality to the past gives
$$
\gamma_\nu(h)=\phi_\nu\gamma_{\nu-1}(h-1)\quad(h\geq1),\qquad V_\nu=\phi_\nu^2V_{\nu-1}+\sigma_\nu^2.
$$
Consequently $\phi_\nu=\gamma_\nu(1)/V_{\nu-1}$ and $\sigma_\nu^2=V_\nu-\gamma_\nu(1)^2/V_{\nu-1}$. Replacing <autocovariance> by matched sample estimates yields seasonal regression estimators. The unit innovation variance is necessary to identify the noise scale separately.
Back to article page
