Seemingly unrelated regressions with heteroskedastic disturbances : some finite sample results

The thesis considers the seemingly unrelated regressions model with heteroskedastic disturbances. It is concerned primarily with a two-equation model in which the heteroskedasticity in the disturbancesarises as a result of the sample data being drawn from two separate populations with unequal variances. Particular attention is given to investigating the finite sample properties of some estimators of the coefficient vector. In Chapter I some instances in which the form of heteroskedasticity considered is appropriate are indicated. These serve to provide the motivation for considering this particular scheme of variance variation. To provide a setting for the subsequent chapters, Chapter II presents an up-to-date survey of the major developments on the estimation of the SUR model. In Chapter III, under the assumption of orthogonal regressors, a unified approach to deriving the exact finite sample second moment matrices of Zellner's restricted and unrestricted residuals variants of the SUR estimators, [...]