Jackknifing the two stage least squares estimator

Abstract

A link is established between jackknifing smooth functions of regression parameter estimators in the general linear model and jackknifing the two stage least squares estimator in the simulta- neous equation model. Hinkley's weighted jackknife is explored and is shown to have a smaller variance than the standard jack- knife estimator in the context of two stage least squares estimation.

Keywords:

• jackknife
• two stage least squares estimation
• weighted jackknife
• Monte Carlo studies