Quandt and Ramsey have suggested an estimator for normal mixtures and switching regressions, which minimizes a sum of squared differences between empirical and theoretical values of the moment generating function. This paper demonstrates how their estimator can be improved by minimizing a generalized sum of squares rather than an ordinary sum of squares. When this is done, more points of evaluation (moments) are unambiguously better than less. Most of the results presented are also applicable to method of moments estimators in general.

This paper considers the asymptotic distribution of forecasts made several time periods in the future. Such forecasts typically arise in the dynamic simulation of an econometric model.

Peter Schmidt, The Algebraic Equivalence of the Oberhofer-Kmenta and Theil-Boot Formulae for the Asymptotic Variance of a Characteristic Root of a Dynamic Econometric Model, Econometrica, Vol. 42, No. 3 (May, 1974), pp. 591-592

Peter Schmidt, On the Difference Between Conditional and Unconditional Asymptotic Distributions of Estimates in Distributed Lag Models with Integer-Valued Parameters, Econometrica, Vol. 41, No. 1 (Jan., 1973), pp. 165-169

assumptions in each individual case, and often a modification of the classical model may become necessary. On the other hand, for any research in which the frontier function is important, the programming methodology is probably a more relevant tool. True, the sampling properties of the programming estimators are mostly unknown. But with the development of modern computer technology, large scale simulation studies can be expected to ascertain the sampling properties under various assumptions. It is the hope of this author that reports about results of such simulation studies will appear in the literature in the near future.