On revenue management and the use of occupancy forecasting error measures

This study aims to draw the attention of the revenue management academic community to inherent problems in forecasting accuracy measurement, and to initiate a critical discussion about forecast quality assessment in hotels. An exhaustive, literature-based set of seventeen forecasting accuracy measures was applied to hotel daily occupancy forecasting data of 2043 pairs of computer and human forecast/actuals, across multiple forecasting horizons. The empirical analysis demonstrates endemic inconsistencies across the accuracy measures, and a plethora of theoretical and practical challenges with regard to total hotel, as well as customer segment level forecast accuracy assessment. The analysis illustrates the difficulty of interpreting conflicting results, as well as issues like level of data aggregation and multiple forecasting horizons. The paper concludes by briefly discussing a more comprehensive approach to hotel forecasting quality assessment framework and serves to warn hotel revenue management academics, practitioners and solution providers against the unconsidered use of accuracy measures.