Classical and Bayesian Inference for Hidden Truncated Pareto Data

• Activity Number: 73
• Type: Contributed
• Date/Time: Sunday, July 31, 2011 : 4:00 PM to 5:50 PM
• Sponsor: Section on Bayesian Statistical Science
• Author(s): Indranil Ghosh*+ and Barry C. Arnold
• Companies: University of California at Riverside and University of California at Riverside
• Keywords: Hidden Truncation ; Bivariate Pareto ; Monte Carlo simulation

Abstract

The Pareto distribution is a simple model for non negative data with a power law probability tail.In fact Pareto's distributions and their close relations and various generalizations provide a very flexible family of distributions that can be used to model specially income distributions as well as a wide variety of other social and economic distributions.Among them Pareto(type(II)) model needs further investigation because together with modeling income distributions it enjoys a greater amount of applicability for the analysis of lifetime data in comparison with other Pareto models.In this paper we focus our attention to estimate the parameters of a bivariate Pareto(Type(II)) distribution when hidden truncation is applied to one of the co variable with the only restriction that the truncation point will be greater than the location parameter of the truncated variable under both the classical and Bayesian paradigm.