Relatore
Falvio Pons
Abstract
In the last two decades, technological development has determined a rapid change of both selling and purchasing dynamics in the accommodation market. On the one hand, modern pricing algorithms are able to determine optimal room rates in response to changes in demand patterns, taking into account the advance booking, the inventory level and the operational cost. This enables hotel managers to perform a dynamic revenue management, adapting their pricing strategy with daily or even higher frequency.
On the other hand, the diffusion of online reservation systems has introduced further complications. In fact, not only these systems facilitate hotel managers in their dynamic pricing practice, but provide both sellers and purchaser a great amount of information about the state of the market. This information can be exploited by customers to optimize their purchase, and by managers to observe the behaviour of their (supposedly) direct competitors almost in real time.
We argue that the superposition of these elements induces two main effects on the market. First, given a location and then a population of competing hotels, we expect to observe a common temporal dynamics, rising from the response to exogenous shocks and to the time-changing nature of demand.
Second, we test if there exists a group of hotels that are market price makers (i.e. display a significant effect of the price dynamics in the considered destination), without being dominant in terms of market share, as would be requested by economic theory. Moreover, we investigate whether such price makers display common features.
In order to assess these two hypotheses, we consider daily best available rates for a panel of 107 hotels in Milan, sampled from January 1st to September 30th, 2016. For each arrival date, the price is given with an advance booking from 0 to 29 days, defining a price trajectory. We investigate the temporal dynamics of the price trajectories by specifying a panel VAR model, including appropriate covariates. Results suggest that only very short- and long-term prices are significant in defining the shape of the price trajectory for the given arrival date, together with the effect of day of week, month, fairs and holidays.
We then exploit pairwise Granger causality to detect relations between hotels and define a direct competition network. We consider network statistics, such as the distributions of in- and out-degree and of the link length between competing hotels, to the aim of defining a simple competition model.
Organizzazione
Alessandra Luati, Silvia Cagnone