Bayesian models for poverty mapping in Thailand

Abstract

Poverty maps are important sources of information for solving social, economic, and environmental problems. Initially, the World Bank used the ELL method to produce poverty maps for used in designing, targeting, prioritizing interventions and allocating the budgets for underdeveloped countries. Even though the ELL method has been shown to have many advantages in poverty mapping, it does not use a survey for the most benefit. Therefore, the Empirical Bayes (EB) method and the hierarchical Bayes (HB) method were proposed in literature. In another aspect, Louis shows that the usual Bayes has a limitation. Therefore, he proposed a new method called constrained Bayes (CB) method. For this reason, we apply this to Empirical Bayes and hierarchical Bayes. This research is divided into two parts. First to study the efficiency of EB and HB compare with the original ELL method by applying to Thai data with the FGT poverty indicators. Second to study the efficiency of constrained Bayes with constrained Empirical Bayes (CEB) and constrained hierarchical Bayes (CHB) by applying to Thai expenditure data