Estimating Parameters in Continuous Univariate Distributions with a Shifted Origin

摘要: SUMMARY A general method of estimating parameters in continuous univariate distributions is proposed. It especially suited to cases where one the an unknown shifted origin. This occurs, for example, three-parameter lognormal, gamma and Weibull models. For such it known that maximum likelihood (ML) estimation can break down because unbounded this lead inconsistent estimators. Properties proposed are described. In particular shown give consistent estimators with asymptotic efficiency equal ML when these exist. Moreover gives consistent, asymptotically efficient situations fails. Examples given including numerical ones showing advantages method.