Likelihood-Based Finite Sample Inference for Synthetic Data from Pareto Model

Abstract

Statistical agencies often publish microdata or synthetic data to protect confidentiality of survey respondents. This is more prevalent in case of income data. In this paper, we develop likelihood-based finite sample inferential methods for a singly imputed synthetic data using plug-in sampling and posterior predictive sampling techniques under Pareto distribution, a well known income distribution. The estimators are constructed based on sufficient statistics and the estimation methods possess desirable properties. For example, the estimators are unbiased and confidence intervals developed are exact. An extensive simulation study is carried out to analyze the performance of the proposed methods.