A Generalized Error Component Model with Heteroscedastic Disturbances.

This paper generalizes the one-way error component model from the homoskedastic to the heteroskedastic case. Unlike P. Mazodier and A. Trognon's (1978) heteroskedastic two-way error component model, an explicit form for V11/2 is obtained that allows the computation of generalized least squares as a simple modification of J. A. Hausman's (1978) procedure for the homoskedastic case. Two methods for estimating the variance components are proposed. The first estimates the variance components based on ordinary least squares and within residuals. The second uses the minimum norm quadratic unbiased estimation procedure suggested by C. R. Rao (1970). These methods are then applied to the estimation of gasoline demand for eighteen OECD countries over the period 1964-78. Copyright 1988 by Economics Department of the University of Pennsylvania and the Osaka University Institute of Social and Economic Research Association.