Bayesian estimation of adult-equivalence scales: An application to Thai expenditure data

Adult-equivalent scales are an important ingredient for assessing living standards and associated aspects such as income inequality and poverty. Various sampling theory procedures, mostly iterative ones, have been suggested in the literature for estimating such scales. However, these estimation procedures do not incorporate prior information such as the nonnegativity of scales and likely inequalities between different scales, and there is concern about an identification problem. In this study Bayesian methodology that includes appropriate prior information is developed and, given appropriate restrictions, no identification problem is evident. The methodology is applied to Thai data involving three expenditure categories, and households distinguished by the numbers in each of three age-gender groups. Both maximum likelihood and the importance sampling technique are used to estimate the adult-equivalent scales for a set of plausible prior information.