We propose an estimator of the Pareto index m, based on the tail behavior of the sample mean. The estimator is strongly consistent and asymptotically normal.
Its behavior is illustrated in a simulation study.