|
A Capital Intensive Approach to the Small Sample Properties of Various Simultaneous Equation Estimators
Robert Summers
Abstract
The small sample properties of five standard simultaneous equation estimating methods are examined by means of the distribution sampling technique. The methods are appraised on the basis of the sampling properties of three different kinds of estimates generated by them: (1) conditional predictions, (2) structural coefficient estimates, and (3) Studentized estimates of structural coefficients. Though one basic model is employed in the investigation, variants used in particular sampling experiments make it possible to compare the methods with respect to bias and efficiency in estimation in the presence of substantial multicollinearity in the redetermined variables and under certain conditions of misspecification. Careful attention is paid to the stochastic variation which necessarily is present in distribution sampling applications.
|
