Some prediction problems concerning samples from the pareto distribution

Previous derivations of predictive distributions within the two-parameter Pareto family have concentrated on individual future observations or on the proportions of future observations which exceed given thresholds, and have typically made use of the relationship between the two-parameter Pareto and the two-parameter exponential models. This relationship is of little help when predictions about sums of future observations are required. In this paper we present a method, which incorporates a missing data approach within a Gibbs sampling routine, to establish predictive densities for such sums.