SARIMA intervention based forecast model for visitor arrivals to Chiang Mai, Thailand

Abstract

The purpose of this research is to implement the forecast model for domestic and international visitor arrivals to Chiang Mai, Thailand using seasonal autoregressive integrated moving average (SARIMA) with intervention analysis. The ADF and extended HEGY tests for the unit root identify that the observed time series are regular and seasonal non-stationary. After differencing of log transformation to the series, the SARIMA model is formulated using monthly data 2000-2007 for the pre-intervention. The residuals obtained from the forecast and secondary data 2008-2013 are assessed with the prior knowledge of various significant crisis events to identify the intervention functions in the forecast model. From the analysis, the violent political turmoil is the major long-term adverse impact on the visitors, whereas the influx of Chinese visitors helps to increase the number of international visitors. The forecasting performance comparison evaluated in terms of the accuracy and reliability indicates that the proposed forecast model outperforms the other existing models for the out-of-sample forecasts. Furthermore, if the government intensifies for solving the internal politics while the provincial administrator can maintain the massive number of Chinese, Chiang Mai will welcome over 10 million visitors and will also generate tourism revenue of about USD 2,400 million in 2018 estimated from the proposed forecast model.