This paper introduces a new class of M-estimators based on generalised empirical likelihood estimation with some auxiliary information available in the sample. The resulting class of estimators is efficient in the sense that it achieves the same asymptotic lower bound as that of the efficient...