Forecasting Chinese International Outbound Tourists: Copula Kink AR-GARCH Model

Pimonpun Boonyasana, Warattaya Chinnakum

Abstract

The aim of this paper is to model and forecast Chinese tourist arrivalsto Thailand, Singapore, and Malaysia. Monthly tourist arrivals from 1999 to 2014are used in the analysis. In this paper, we propose Kink AR(m)-GARCH(p,q)model that combine the classical GARCH model of Bollerslev (1986)[1] with theKink model of Chan and Tsay (1998). In additional, we assume that there aredependence between growth rate of tourist arrivals to Thailand, Singapore, andMalysia, from China. Copula approach was employed to capture these dependency. Therefore, Copula-base Kink AR(m)-GARCH(p,q) was used in this study. According to minimizing The results show that T-Copula 2-regimes Kink AR(1)-GARCH(1,1) model with normal, student-t, and skewed student-t error distributions, delivers the most accurate predictions.