A note on the double k-class estimator in simultaneous equations

Dwivedi and Srivastava (1984, DS) studied the exact finite sample properties of Nagar’s (1962) double k-class estimator as continuous functions of its two characterizing scalars k1 and k2, and provided guidelines for their choice in empirical work. In this note we show that the empirical guidelines provided by DS are not entirely valid since they did not explore the complete range of the relevant parameter space in their numerical evaluations. We find that the optimal values of k2 leading to unbiased and mean squared error (MSE) minimizing double k-class estimators are not symmetric with respect to the sign of the product ρω12, where ρ is the correlation coefficient between the structural and reduced form errors, and w12 is the covariance between the unrestricted reduced form errors. Specifically, when ρω12 is positive,the optimal value of k2 is generally positive and greater than k1, which partly explains the superior performance of Zellner’s (1998) Bayesian Method of Moments (BMOM) and Extended MELO estimators reported in Tsurumi (1990).