Estimation Using Marginal Competitor Sales Information

An abiding preoccupation for firms is how customers value their product compared to the competitor's. This is difficult to quantify and estimate from data as even though prices are public information, competitors' sales are typically unobservable. There are some industries however, most prominently the hotel industry, where marginal aggregated information about competitor sales can be obtained through third-party information brokers. In the hotel industry these reports (called STR reports) are widely subscribed; a hotel can participate by reporting its sales information and in turn subscribe to get the marginal sales information of its competitor set, aggregated across groups and the lengths-of-stay. Such data however is not widely incorporated into revenue management (RM) estimation, possibly for the lack of robust models and methodologies to do so. In this paper we tackle this estimation problem under a simple market-share model, focusing on the hotel industry, and show how to overcome the following significant challenges: (i) the fact that competitor's data is aggregated across multiple lengths-of-stay with distinct demands (ii) inability to observe no-purchasers, i.e. those who purchase neither ours nor the competitor's products, and finally, (iii) the competitor makes private sales to groups before the retail sales period. Thus even the competitor's capacity is usually unobservable. Using Monte-Carlo simulation and a model of generalized Nash competition, we first test our procedure on synthetic data; our method nearly recovers the true parameters in all cases. We then apply it to real hotel bookings data, comparing with alternate estimation methods from the network tomography and RM literature.