Forecasting long-haul tourism demand for Hong Kong using error correction models
Abstract:Forecasting accuracy is particularly important when forecasting tourism demand on account of the perishable nature of the product. This study compares a range of forecasting models in the context of predicting annual tourist flows into Hong Kong from the major long-haul markets of the US, the UK, Germany and major short-haul markets of China, Japan and Taiwan. Econometric forecasting models considered included Error Correction Models (ECMs) based on Permanent Income-Life Cycle (PI-LC) hypothesis (PI-LC ECM) and alternative cointegration approaches: Engle and Granger (1987), Johansen (1988), and Ordinary Least Square (OLS) approaches. Both Autoregressive Integrated Moving Average (ARIMA) and no change model (hereafter NAIVE) models are used as a benchmark time series model for accuracy comparisons. It was hypothesized that PI-LC ECM is a better forecasting model particularly for long-haul tourism demand. The objective of this article is to investigate whether the application of PI-LC ECM could improve the forecasting performance of econometric models relative to time series models. The forecasting results indicate that the PI-LC ECM based on the Engle-Granger (1987) approach produces more accurate forecasts than other alternative forecasting models for all long-haul markets based on Mean Absolute Error (MAE) and Root Mean Square Error (RMSE) criteria. Overall, PI-LC ECMs produce better forecasts of tourism demand than the OLS, ARIMA and NAIVE models for all origin markets and all time horizons.
Document Type: Research Article
Affiliations: Division of Commerce, City University of Hong Kong, Tat Chee Avenue, Kowloon Tong, Kowloon, Hong Kong, China
Publication date: February 1, 2011