ABSTRACT
This chapter examines the properties of various estimators in a multiple regression model when regression coefficients are purely random. It discusses the asymptotic properties of some of the commonly used estimators. The chapter also discusses the Monte Carlo results regarding the small sample performance of these and other estimators of the means and variances of purely random coefficients. It shows that the sampling error of the operational generalised least squares estimator (OGLS) estimator, that is the OGLS estimator is a consistent estimator of. The chapter explores that the Hildreth and Houck (HH) estimator of may be obtained from the regression equation where the disturbance is vector such that tends to zero. The summary conclusions are: The determinant value of the moment matrix of the asymptotic distribution underestimates the determinant value of the finite sample approximate moment matrix of the estimator for small samples and the size of underestimation of moment matrix of varies inversely.
