Error Components Regression Models and Their Applications

In this paper, we have developed an operational method for estimating error components regression models when the variance- covariance matrix of the disturbance terms is unkown. Monte Carlo Studies were conducted to compare the relative efficiency of the pooled estimator obtained by this procedure to (a) an ordinary least sources estimator based on data aggregated over time, (b) the covariance estimator, and (d) a generalized least squares estimator based on a known variance-covariance matrix. For T small and large p, this estimator definitely performs better than the other estimators which are also based on an estimated value of the variance-covariance matrix of the disturbances. For p small and large T it compares equally well with other estimators.