CMStatistics 2016: Start Registration
View Submission - CFE
Title: GARCH model for income time series data and forecasting income inequality Authors:  Haruhisa Nishino - Hiroshima University (Japan) [presenting]
Abstract: Income inequality is a great issue also in the Japanese economy as well as the world economy. We take an approach by use of a parametric model for analyzing income inequality. Lognormal distribution, which is used, is better fitted to Japanese income data and useful for extracting inequality from income data because its scale parameter only represents inequality. We propose GARCH models including income inequality for income time series data. The GARCH model is suitable for modelling the scale parameter of lognormal distribution. A joint distribution from selected order statistics enables us to construct a quasi-likelihood of the GARCH model from quantile income data. The proposed model has an inequality structure and a time series structure. It is useful for analyzing persistent income inequality and forecasting income inequality. The models are estimated from Japanese quantile income data (Family Income and Expenditure Survey by Ministry of Internal Affairs and Communications, Statistics Bureau, Japan) and the various GARCH models are compared and examined from the point of view of forecasting.