Time series forecasting of post-war tourism prospects for Sri Lanka

Abstract

Tourism plays a big role in the development of a country in terms of economics as it is one of the biggest and fastest-growing economic sectors in the world. It accounts for a large part of Gross Domestic Product of any country through Foreign Exchange. This study focused on international tourist arrivals to Sri Lanka. In the past, nearly three decades, Sri Lanka had to face conflict within the country. Tourists had less interest of visiting Sri Lanka, mainly due to the uncertainty of security. Nevertheless, the internal conflict is over and tourist arrivals have dramatically increased over last six years. The aim of this study is to investigate the impact of internal conflict in Sri Lanka for tourist arrivals by splitting the entire time frame by before and after the conflict as two windows. Further this study discusses the factors which are influenced by tourism in Sri Lanka. The data for the study is extracted from the annual reports of the Sri Lanka Tourism Development Authority. Time series models are developed in two separate time windows by using the methods: HoltWinters’ Exponential Smoothing, Seasonal Autoregressive Integrated Moving Average (ARIMA) modeling, State Space modeling and Dynamic Transfer Function modeling. All necessary tests are carried out for model development, diagnostic checking and forecast. In the empirical study, behavior of arrivals with its trend and seasonal patterns are analyzed, best models are developed based on the accuracy of fitted models in terms of Mean Absolute Percentage Error (MAPE) values and the impact ofthe factors influenced by tourism are deeply discussed. MAPE values for the recommended models for after the conflict arc less than 7%. In both windows, Seasonal ARIMA method performs the best. Moreover it is estimated by ex-post forecast that, 2.085 million international tourist arrivals can be expected in the year 2016.