Bootstrap Approach to Comparison of Alternative Methods of Parameter Estimation of a Simultaneous Equation Model

Abstract

A bootstrap simulation approach was used to generate values for endogenous variables of a simultaneous equation model popularly known as Keynesian Model of Income Determination. Three sample sizes 20, 30 and 40 each replicated 10, 20 and 30 times were
considered. Four different estimation techniques: Ordinary Least Square (OLS); Indirect Least Square (ILS); Two-Stage Least Square (2SLS) and Full Information Maximum Likelihood (FIML) methods were employed to estimate the parameters of the model. The
estimators were then evaluated using the average parameter estimates; absolute bias of the estimates and the root mean square error of the estimates. The result shows that generally, ILS provided the best estimates.

Keywords: Bootstrap, endogenous, exogenous, least squares, maximum likelihood.

African Research Review Vol. 2 (3) 2008: pp. 51-61