A0502
Title: A statistical approach to evaluate last minute pricing decisions in the online hotel market
Authors: Luca Vincenzo Ballestra - Alma Mater Studiorum University of Bologna (Italy)
Enzo DInnocenzo - VU University Amsterdam (Netherlands)
Andrea Guizzardi - Alma Mater Studiorum University of Bologna (Italy) [presenting]
Abstract: A nonlinear statistical framework is proposed for studying the last-minute pricing decisions of hotels active in the online market. In particular, the last-minute rate is specified as a linear function of the early booking rate and a shock term. The latter is regarded as the combined effect of the hoteliers forecasting error about the pick-up curve and their last-minute pricing tactics. We connect the parameters of the shock distribution to the revenue managers' practices, modeling location, scale, skewness and kurtosis with a dynamic score-driven approach. To overcome possible issues of endogeneity, a nonlinear instrumental variable estimator is employed. An empirical analysis is performed where we leverage a large dataset obtained by scraping information that is publicly available on the internet. Results show that the error term is accurately described by a skew-$t$ distribution, rather than by a Gaussian specification. Moreover, the score-driven model turns out to be very suitable for capturing the complex nonlinear behavior of online everyday pricing decisions. Actually, the proposed approach is a reliable and transparent tool to assess the online pricing behavior of any hotel that publishes rates on the internet.
