Bayesian estimation of Persistent Income Inequality by Lognormal Stochastic Volatility Model

Abstract

We estimate inequality including Gini coefficients using a lognormal parametric model for an investigation of persistent inequality. The asymptotic theory of selected order statistics enables us to construct a linear model based on grouped data. We extend the linear model to a dynamic model in terms of a stochastic volatility (SV) model. Using Japanese data we estimate the SV model by the Markov chain Monte Carlo (MCMC) method and exploit a model comparison to choose a best model, concluding that the model with SV is better fitted to the data than the model without SV. It indicates the persistent inequality.