Modeling price-sensitive demand in turbulent times: an application to continuous pricing

Pricing drives demand for service industries such as air transport, hotels, and car rentals. To optimise the price, firms have to predict real-time customer demand at the micro level and optimise the price. This paper contributes to revenue management by introducing a nonparametric statistical approach to predict price-sensitive demand and its application to continuous pricing. Continuous pricing lets service companies maximise revenue by using customers’ willingness to pay. However, it requires accurate demand estimations, particularly of customers’ price sensitivity. This paper introduces an augmented generalised additive model to estimate price sensitivity, which identifies substantial variations in price sensitivity, exceeds the predictive performance of state-of-the-art alternatives, and controls for price endogeneity. In addition, the demand model has variable price derivatives enabling continuous pricing. The proposed approach offers a simple and efficient way to implement continuous pricing with a closed-form solution. Our research also highlights the relevance of considering the problem of price endogeneity when estimating price-sensitive demand based on observations from prior pricing decisions. We demonstrate how continuous pricing is applied using empirical airline ticket data. We document a field study, which shows a revenue increase of 6% on average, and outline how the approach applies to turbulent market conditions caused by the COVID-19 pandemic, the surge in inflation since mid-2021, and the start of the Ukraine war in April 2022.