Developing a two way error component estimation model with disturbances following a special autoregressive (4) for quarterly data

This paper provides an estimation method for a two way error component regression model where the time-varying disturbances are serially correlated, following a special AR (4) process for quarterly data. The variance-covariance matrix of the compound error terms and its spectral decomposition are also derived, allowing the computation of the Generalized Least Square (GLS) estimates and residuals. The Best Quadratic Unbiased (BQU) Estimates of the variance components are proposed, as well as estimates of all parameters involved in the resulting feasible GLS method.