Dependence issues in tourism demand modeling and forecasting : within and between dependence

Abstract

Tourism forecasting has long been an attractive topic in the tourism demand literature because of the undoubted significance of such studies for policy purposes. Many researchers have applied or modified various techniques routinely used in business for computing tourism forecasts and increasing forecast accuracy. Among these techniques, linear models are popular as they are computationally straightforward and can be easily applied in practice. However, linear relationship is just one of the numerous dependence structures, and the misspecification of dependence structures can cause over- or under-estimation problem.

This thesis proposes the copula technique to address this issue. The research focuses on international tourism demand for Singapore and applies the copula method to examine tourism flows/arrivals to Singapore from 1995 to 2013. The empirical analysis of this thesis covers not only the serial dependence (labeled as Within Dependence) of individual tourism flows to Singapore, but also the dependence between different tourism flows (labeled as Between Dependence). Specifically, I conduct four studies to advance knowledge of dependence structure in tourism demand modeling and make a contribution to tourism forecasting with the use of the copula method. Study 1 and 2 analyze Within Dependence. The simple two-dimensional model is initially used to examine the Within Dependence of tourism demand and its first lag in Study 1. Study 2 investigates the sophisticated high-dimensional cases and includes higher order lags of tourism demand. The third study examines whether the dependence between tourism flows exists, to shed light on the Between Dependence and contribute to the joint forecast of tourism flows. In the last study, I take both the Within and Between Dependences into account, and develop a comprehensive copula-WB framework for forecasting.

By analyzing the tourism flows to Singapore, the empirical results of the studies reveal the different serial dependence structures of tourism flows per se and the dependence structures between flows. Additionally, the studies show that in contrast to the AR, ARIMA, SARIMA, ARDL and ECM models, the copula method is flexible in specifying dependence structure and can improve tourism demand forecasting.