Small sample properties of an inequality pretest ridge regression estimator in a misspecified linear model

This paper derives a pre-test ridge regression estimator of the regression coefficients in the presence of an inequality constraint and examines the small sample properties of the estimator in a model which is mis-specified through the omission of relevant regressors. The numerical results show that the pre-test estimator performs at least as well as the inequality constraint ridge regression estimator in terms of the risk properties in most of the cases. In addition, a bootstrap procedure is proposed for estimating the bias and the mean squared error (MSE) of the pre-test estimator, and the numerical evaluations demonstrate that the bootstrap method is a good method to estimate the small sample properties.