Estimating bivariate errors-in-variables models with instrumental variables

Abstract

The bivariate errors-in-variables model is considered with a new choice of instrumental variables. A new estimate of the regression coefficient, based on these instrumental variables, has attractive properties over the ordinary least square estimator and Wald's estimator. Monte Carlo simulation studies indicate that the proposed estimator is than its competitors over a range of plausible parameter values, where better is defined in terms of mean square error and Pitman's nearness criterion. Finally, we use actual data from an assessment equity study to show that our proposed method produces estimators that, appear superior to those produced by the ordinary least square estimator and Wald's estimator