An evaluation of the bid price and nested network revenue management allocation methods. (January 2018)

• Record Type: Journal Article
• Authors: Pimentel, Victor; Aizezikali, Aishajiang; Baker, Tim
• Is Part Of: Computers & industrial engineering. Volume 115(2018)
• Page Start: 100
• Page End: 108
• Publication Date: 2018-01
• Subjects: Simulation -- Heuristics -- Nonlinear programming; Engineering -- Data processing -- Periodicals; Industrial engineering -- Periodicals; 620.00285
• DOI: 10.1016/j.cie.2017.10.026
• Languages: English
• ISSNs: 0360-8352

Abstract

Highlights: We compare the bid price and nested network methods. We simulate a hotel reservation system for transient customers. The nested network approach outperforms the bid price approach. Our new approaches improve net revenue in a worst-case evaluation. It would be worthwhile for industry to implement simultaneous approaches. Abstract: We compare the revenue generating capabilities of the bid price allocation method and the nested network method in hotel revenue management. Revenue maximization is achieved by an optimal allocation of assets across market segments, subject to constraints such as overbooking limit and the cross-elasticity of competitors' pricing. Using a simulation model of a large hotel's reservation system, validated by Marriott hotels, we find that the nested network method outperforms the bid price method, and, on average, leads to an improvement of 6% in revenue in the worst-case scenarios across operating environments. This improvement is 3.6% when restricted to cases in which overbooking and allocation are performed simultaneously. In no operating environment is the improvement less than 2%. Since the bid price method is, by far, the most commonly used allocation method in practice, these results indicate that hotels should consider switching to the nested network method. This change is feasible because (1) most hotels already have in place the core optimization system required to execute the nested network method, and (2) the nested network method converges to optimality in less than two minutes for most realistically sized problems, as we demonstrate.