Seasonality has attracted considerable interest in empirical tourism research and forecasting. However, the analysis of such recurring phenomenon is sparse in hospitality research, with very few studies to date having analysed seasonal unit roots prior to forecasting guest nights for the tourist-lodging industry. While pricing and other strategies have been implemented to extract greater revenue in the hospitality and tourism industry, it is also essential to focus the application of revenue or yield management based on seasonal demand analysis and forecasting. This paper examines the seasonality of hotel–motel room night occupancy patterns in New Zealand using monthly time series from 1997 to 2007. The presence of seasonal unit roots is detected using the Hylleberg, Engel, Granger and Yoo (HEGY) procedures. When the Box–Jenkins model of the HEGY-based transformed series is used to forecast hotel–motel room nights, its forecast performance is worse than that of the 12 differenced SARMA(2, 2)(0, 2) 12 model and the ARMA(2, 2) model for the original series.
