Contrasting revised empirical likelihood and its competitors

Abstract

The Minimum Empirical Divergence methods operate within the Empirical Estimating Equations (E3) approach to estimation and inference. Consequently, existence of E3 estimators and tests is data dependent and subject to the Empty Set Problem (ESP). As one option to avoid ESP we suggest the Revised Empirical Likelihood (ReEL) method, which operates on the level of Estimating Equations. The Euclidean Empirical Likelihood (EEL), which allows for negative weights in E3, also avoids ESP. To indicate the relative sampling performance, we compare the small sample performance of ReEL and EEL in an example from the literature.