How do Less Advanced Forecasting Methods Perform on Weekly RevPAR in Different Forecasting Horizons Following the Recession?

Abstract

The purpose of this study is to examine the performance of three smoothing methods on forecasting weekly revenue per Available room (revPAr) following the recent recession in comparison to more sophisticated time series forecasting methods. The results of this study show that simpler methods perform better. Simple moving Average and Single Exponential Smoothing outperformed Autoregressive Integrated moving Average and Artificial Neural Networks in all 10 of 5-week forecasting horizons, which suggests accurate weekly revPAr forecasting in both short and long term can be achieved with simple, easy-to-learn, yet effective methods. The findings of the study are expected to not only contribute to the limited literature of revPAr forecasting, but also provide practitioners with empirical evidence for selecting appropriate time series forecasting methods for weekly revPAr forecasting in different time horizons, particularly following an economic downturn.

Forecasting tourist arrivals using origin country macroeconomics

“…It should be noted that a remarkable number of studies concentrate on the impact of macroeconomic indicators of origin countries on tourism demand at destination countries. For its most part though, literature in this field has 1 See, inter alia, Kulendran and Wilson, 2000;Cho, 2001;Lim and McAleer, 2001a;Goh and Law, 2002;Cho, 2003;Kulendran and Witt, 2003;Chen, 2005;Vu and Turner, 2005;Kim and Moosa, 2005;Vu and Turner, 2006;Wong et al, 2007;Coshall, 2009;Santos, 2009;Brida and Risso, 2011;Gounopoulos et al, 2012;Zheng et al, 2012;Wan et al, 2013. 2 See for instance, Kulendran and Wilson, 2000;Song and Witt, 2000;Kulendran and Witt, 2003;Song and Witt, 2006;Wong et al, 2006;Wong et al, 2007.…”

“…3 See indicatively, Song et al, 2003;Wong et al, 2007. 4 See, for example, Lim and McAleer, 2001b;Chen, 2005;Vu and Turner, 2005;Yu et al, 2007;Zheng et al, 2012. focused on demand factors and especially income (see, among others, Nicolau and Mas, 2005a,b;Wang et al, 2006;Rudez, 2008;Lim et al, 2009;Alegre et al, 2010). This is mainly due to the fact that tourism is regarded as a luxury good and as such, it is expected to be heavily dependent on income (Chatziantoniou et al, 2013;Wang, 2014).…”

Occupancy forecasting methods and the use of expert judgement in hotel revenue management

“…The time series models studied in the academic literature about hospitality and tourism include autoregressive integrated moving average (ARIMA) model, autoregressive distributed lag model (ADLM), error correction model (ECM), vector autoregressive (VAR) model (Wong, Song, Witt, & Wu, 2007) Box-Jenkins procedure (ARIMA Models), and smoothing methods such as simple moving average, single exponential smoothing, and Holt-Winters (Zheng, Bloom, Wang, & Schrier, 2012). Many hotel revenue management systems rely on the approaches of exponential smoothing (Holt-Winters), moving average methods (simple and weighted), or linear regression to forecast demand based on historical arrivals (Weatherford & Kimes, 2003).…”

“…Weatherford, Kimes, and Scott (2001) also tested the accuracy of aggregated and disaggregated forecasting methods and found that the disaggregated forecasts outperformed the various aggregated methods. Zheng, Bloom, Wang, and Schrier (2012) found in the forecasting of RevPar, simple moving average and single exponential smoothing methods outperformed ARIMA and artificial neural networks. A review of recent literature suggests it is very challenging to identify a "best" forecasting method (Song & Li, 2008).…”

“…The ability to accurately forecast arrivals to a tourist destination is important to minimize the financial and environmental costs of excess capacity or the opportunity costs of unfulfilled demand (Chu, 2011). The time series models studied in the academic literature about hospitality and tourism include autoregressive integrated moving average (ARIMA) model, autoregressive distributed lag model (ADLM), error correction model (ECM), vector autoregressive (VAR) model (Wong, Song, Witt, & Wu, 2007), Box-Jenkins Procedure (ARIMA models), and smoothing methods such as simple moving average, single exponential smoothing, and Holt-Winters (Zheng, Bloom, Wang, & Schrier, 2012).…”